Machine Learning in Automated Text Categorization

2009

This paper presents an approach for categorizing documents according to their implicit locational relevance. We report a thorough evaluation of several classifiers designed for this task, built by using support vector machines with multiple alternatives for feature vectors. Experimental results show that using feature vectors that combine document terms and URL n-grams, with simple features related to the locality of the document (e.g. total count of place references) leads to high accuracy values. The paper also discusses how the proposed categorization approach can be used to help improve tasks such as document retrieval or online contextual advertisement.

2003

In the Internet era, huge amounts of data are available to everybody, in every place and at any moment. Searching for relevant information can be overwhelming, thus contributing to the user’s sense of information overload. Building systems for assisting users in this task is often complicated by the difficulty in articulating user interests in a structured form – a profile – to be used for searching. Machine learning methods offer a promising approach to solve this problem. Our research focuses on supervised methods for learning user profiles which are predictively accurate and comprehensible. The main goal of this paper is the comparison of two different approaches for inducing user profiles, respectively based on Inductive Logic Programming (ILP) and probabilistic methods. An experimental session has been carried out to compare the effectiveness of these methods in terms of classification accuracy, learning and classification time, when coping with the task of learning profiles from textual book descriptions rated by real users according to their tastes.

International Journal on Artificial Intelligence Tools, 2007

The increasing number of unsolicited e-mail messages (spam) reveals the need for the development of reliable anti-spam filters. The vast majority of content-based techniques rely on word-based representation of messages. Such approaches require reliable tokenizers for detecting the token boundaries. As a consequence, a common practice of spammers is to attempt to confuse tokenizers using unexpected punctuation marks or special characters within the message. In this paper we explore an alternative low-level representation based on character n-grams which avoids the use of tokenizers and other language-dependent tools. Based on experiments on two well-known benchmark corpora and a variety of evaluation measures, we show that character n-grams are more reliable features than word-tokens despite the fact that they increase the dimensionality of the problem. Moreover, we propose a method for extracting variable-length n-grams which produces optimal classifiers among the examined models under cost-sensitive evaluation.

This paper uses Systemic Functional Linguistic (SFL) theory as a basis for extracting semantic features of documents. We focus on the pronominal and determination system and the role it plays in constructing interpersonal distance. By using a hierarchical system model that represents the author's language choices, it is possible to construct a rich and informative feature representation.

2009

The bias in the news media is an inherent flaw of the news production process. The resulting bias often causes a sharp increase in political polarization and in the cost of conflict on social issues such as Iraq war. It is very difficult, if not impossible, for readers to have penetrating views on realities against such bias. This paper presents NewsCube, a novel Internet news service aiming at mitigating the effect of media bias. NewsCube automatically creates and promptly provides readers with multiple classified viewpoints on a news event of interest. As such, it effectively helps readers understand a fact from a plural of viewpoints and formulate their own, more balanced viewpoints. While media bias problem has been studied extensively in communications and social sciences, our work is the first to develop a news service as a solution and study its effect. We discuss the effect of the service through various user studies.

2009

This paper presents an architecture which applies document similarity measures to the documentation produced during the phases of software development in order to generate recommendations of process and people metrics for similar projects. The application makes a judgment of similarity of the Service Provision Offer (SPO) document of a new proposed project to a collection of Project History Documents (PHD), stored in a repository of unstructured texts. The process is carried out in three stages: firstly, clustering of the Offer document with the set of PHDs which are most similar to it; this provides the initial indication of whether similar previous projects exist, and signifies similarity. Secondly, determination of which PHD in the set is most comparable with the Offer document, based on various parameters: project effort, project duration (time), project resources (members/size of team), costs, and sector(s) involved, indicating comparability of projects. The comparable parameters are extracted using the GATE Natural Language Processing architecture. Lastly, a recommendation of metrics for the new project is made, which is based on the transferability of the metrics of the most similar and comparable PHD extracted, here referred to as recommendation.

2003

The k-Nearest-Neighbours (kNN) is a simple but effective method for classification. The major drawbacks with respect to kNN are (1) its low efficiency – being a lazy learning method prohibits it in many applications such as dynamic web mining for a large repository, and (2) its dependency on the selection of a “good value” for k. In this paper, we propose a novel kNN type method for classification that is aimed at overcoming these shortcomings. Our method constructs a kNN model for the data, which replaces the data to serve as the basis of classification. The value of k is automatically determined, is varied for different data, and is optimal in terms of classification accuracy. The construction of the model reduces the dependency on k and makes classification faster. Experiments were carried out on some public datasets collected from the UCI machine learning repository in order to test our method. The experimental results show that the kNN based model compares well with C5.0 and kNN in terms of classification accuracy, but is more efficient than the standard kNN.

ACM Transactions on Database Systems, 2007

Text is ubiquitous and, not surprisingly, many important applications rely on textual data for a variety of tasks. As a notable example, information extraction applications derive structured relations from unstructured text; as another example, focused crawlers explore the web to locate pages about specific topics. Execution plans for text-centric tasks follow two general paradigms for processing a text database: either we can scan, or "crawl," the text database or, alternatively, we can exploit search engine indexes and retrieve the documents of interest via carefully crafted queries constructed in task-specific ways. The choice between crawl-and query-based execution plans can have a substantial impact on both execution time and output "completeness" (e.g., in terms of recall). Nevertheless, this choice is typically ad-hoc and based on heuristics or plain intuition. In this article, we present fundamental building blocks to make the choice of execution plans for text-centric tasks in an informed, cost-based way. Towards this goal, we show how to analyze query-and crawl-based plans in terms of both execution time and output completeness. We adapt results from random-graph theory and statistics to develop a rigorous cost model for the execution plans. Our cost model reflects the fact that the performance of the plans depends on fundamental task-specific properties of the underlying text databases. We identify these properties and present efficient techniques for estimating the associated parameters of the cost model. We also present two optimization approaches for text-centric tasks that rely on the costmodel parameters and select efficient execution plans. Overall, our optimization approaches help build efficient execution plans for a task, resulting in significant efficiency and output completeness benefits. We complement our results with a large-scale experimental evaluation for three important text-centric tasks and over multiple real-life data sets.

Frontiers of Computer Science in China, 2007

Research and Development Program (HTRDP) sponsored a series of technology evaluations on Chinese information processing and intelligent human-machine interface, which is called HTRDP evaluations, or "863" evaluations in brief. This paper introduces the HTRDP evaluations in detail. The general information of the HTRDP evaluation is presented first, including the history, the concerned technology categories, the organizer, the participants, and the procedure, etc. Then the evaluations on each technology are described in detail respectively, covering Chinese word segmentation, machine translation, acoustic speech recognition, text to speech, text summarization, text categorization, information retrieval, character recognition, and face detection and recognition. For the evaluations on each technology categories, the history, the evaluation tasks, the data, the evaluation method, etc., are given. The last section concludes the paper and discusses possible future work.

Journal of The American Society for Information Science and Technology, 2004

Text Categorization is the process of assigning documents to a set of previously fixed categories. A lot of research is going on with the goal of automating this time-consuming task. Several different algorithms have been applied, and Support Vector Machines (SVM) have shown very good results. In this report, we try to prove that a previous filtering of the words used by SVM in the classification can improve the overall performance. This hypothesis is systematically tested with three different measures of word relevance, on two different corpus (one of them considered in three different splits), and with both local and global vocabularies. The results show that filtering significantly improves the recall of the method, and that also has the effect of significantly improving the overall performance.

2004

In this paper we present a novel algorithm for document clustering. This approach is based on distributional clustering where subject related words, which have a narrow context, are identified to form meta-tags for that subject. These contextual words form the basis for creating thematic clusters of documents. In a similar fashion to other research papers on document clustering, we analyze the quality of this approach with respect to document categorization problems and show it to outperform the information theoretic method of sequential information bottleneck.

Applied Artificial Intelligence, 2008

Journal of Internet Services and Applications, 2012

The use of web services in enterprise applications is quickly increasing. In a web services environment, providers supply a set of services for consumers. However, although web services are being used in business-critical environments, there are no practical means to test or compare their robustness to invalid and malicious inputs. In fact, client applications are typically developed with the assumption that the services being used are robust, which is not always the case. Robustness failures in such environments are particularly dangerous, as they may originate vulnerabilities that can be maliciously exploited, with severe consequences for the systems under attack. This paper addresses the problem of robustness testing in web services environments. The proposed approach is based on a set of robustness tests (including both malicious and non-malicious invalid call parameters) that is used to discover programming and design errors. This approach, useful for both service providers and consumers, is demonstrated by two sets of experiments, showing respectively the use of web services robustness testing from the consumer and the provider points of view. The experiments comprise the robustness testing of 1204 web service operations publicly available in the Internet and of 29 home-implemented services, including two different implementations of the web services specified by the standard TPC-App performance benchmark. Results show that many web services are deployed with critical robustness problems and that robustness testing is an effective approach to improve services quality.

Lecture Notes in Computer Science, 2004

Knowledge management solutions relying on central repositories sometimes have not met expectations, since users often create knowledge adhoc using their individual vocabulary and using their own decentral IT infrastructure (e.g., their laptop). To improve knowledge management for such decentralized and individualized knowledge work, it is necessary to, first, provide a corresponding IT infrastructure and to, second, deal with the harmonization of different vocabularies/ontologies. In this paper, we briefly sketch the technical peer-to-peer platform that we have built, but then we focus on the harmonization of the participating ontologies. Thereby, the objective of this harmonization is to avoid the worst incongruencies by having users share a core ontology that they can expand for local use at their will and individual needs. The task that then needs to be solved is one of distributed, loosely-controlled and evolving engineering of ontologies. We have performed along these lines. To support the ontology engineering process in the case study we have furthermore extended the existing ontology engineering environment, OntoEdit. The case study process and the extended tool are presented in this paper.

Lecture Notes in Computer Science, 2009

We present a model of user affect to recognize multiple user emotions during interaction with an educational computer game. Our model deals with the high level of uncertainty involved in recognizing a variety of user emotions by probabilistically combining information on both the causes and effects of emotional reactions. In previous work, we presented the performance and limitations of the model when using only causal information. In this paper, we discuss the addition of diagnostic information on user affective valence detected via an EMG sensor, and present an evaluation of the resulting model.

Group Decision and Negotiation, 2006

Various combination of Natural Language Processing and Machine Learning methods offer ample opportunities wherever texts are an important element of an application or a research area. Such methods discover patterns and regularities in the data, seek generalization and in effect learn new knowledge. We have employed such methods in learning from a large amount of textual data. Our application is electronic negotiations. The genre of texts found in electronic negotiations may seem limited. It is an important research question whether our methods and findings apply equally well to texts that come from face-to-face negotiations. In order to confirm such more general applicability, we have analyzed comparable collections of texts from electronic and face-to-face negotiations. We present our findings on the extent of similarity between these two related but distinct genres. In this study we have analyzed similarities in the text data of electronic and face-to-face negotiations. The results show that -in certain conditions -vocabulary richness, language complexity and text predictability are similar.

Lecture Notes in Computer Science, 2009

Data pre-processing is an important topic in Text Classification (TC). It aims to convert the original textual data in a data-mining-ready structure, where the most significant text-features that serve to differentiate between textcategories are identified. Broadly speaking, textual data pre-processing techniques can be divided into three groups: (i) linguistic, (ii) statistical, and (iii) hybrid (i) & (ii). With regard to language-independent TC, our study relates to the statistical aspect only. The nature of textual data pre-processing includes: Document-base Representation (DR) and Feature Selection (FS). In this paper, we propose a hybrid statistical FS approach that integrates two existing (statistical FS) techniques, DIAAF (Darmstadt Indexing Approach Association Factor) and GSSC (Galavotti⋅Sebastiani⋅Simi Coefficient). Our proposed approach is presented under a statistical "bag of phrases" DR setting. The experimental results, based on the well-established associative text classification approach, demonstrate that our proposed technique outperforms existing mechanisms with respect to the accuracy of classification.

Lecture Notes in Computer Science, 2010

                                                          

Lecture Notes in Computer Science, 2012

In this paper, we proposed a new method for Text Categorization based on fuzzy soft set theory so called fuzzy soft set classifier (FSSC). We use fuzzy soft set representation that derived from the bag-of-words representation and define each term as a distinct word in the set of words of the document collection. The FSSC categorize each document by using fuzzy c-means formula for classification, and use fuzzy soft set similarity similarity to measure distance between two documents. We perform the experiments with the standard Reuters-21578 dataset, and using tree kind of weigthing such as boolean, term frequency, and term frequency-invert document frequency to compare the performance of FSSC with others four classifier such as kNN, Bayesian, Rocchio, and SVM. We are using precision, recall, F-measure, retun-size, and the running time as a performance evaluation. Result shown that there is no absolute winner. The FSSC has precision, recall, and F-measure lower than SVM, and kNN but FSSC can work faster than both. When compared with the Bayesian and Rocchio, the FSSC works more slowly but has a higher precision and Fmeasure.

Journal of Ambient Intelligence and Humanized Computing, 2013

Your article is protected by copyright and all rights are held exclusively by Springer-Verlag. This e-offprint is for personal use only and shall not be self-archived in electronic repositories. If you wish to self-archive your work, please use the accepted author's version for posting to your own website or your institution's repository. You may further deposit the accepted author's version on a funder's repository at a funder's request, provided it is not made publicly available until 12 months after publication.

Applied Spectroscopy, 2010

A multi-group classifier based on the support vector machine (SVM) has been developed for use with a library of 48 456 spectra measured by nearinfrared reflection microscopy (NIRM) on 227 samples representing 26 animal feed ingredients and 4 possible contaminants of animal origin. The performance of the classifier was assessed by a five-fold cross-validation, dividing at the sample level. Although the overall proportion of misclassifications was 27%, almost all of these involved the confusion of pairs of similar ingredients of vegetable origin. Such confusions are unimportant in the context of the intended use of the library, which is the detection of banned ingredients in animal feed. The error rate in discrimination between permitted and banned ingredients was just 0.17%. The performance of the SVM classifier was substantially better than that of the K-nearest-neighbors method employed in previous work with the same library, for which the comparable error rates are 36% overall and 0.39% for permitted versus banned ingredients.

Journal of the Association for Information Science and Technology, 2015

Recognizing negative and speculative information is highly relevant for sentiment analysis. This paper presents a machine learning approach to automatically detecting this kind of information in the review domain. The resulting system works in two steps: in the first pass, negation/speculation cues are identified and in the second phase, the full scope of these cues is determined. The system is trained and evaluated on the Simon Fraser University Review corpus, which is extensively used in opinion mining. The results show how the proposed method outstrips the baseline by as much as roughly 20% in the negation cue detection and around 13% in the scope recognition, both in terms of F 1 . In speculation, the performance obtained in the cue prediction phase is close to that obtained by a human rater carrying out the same task. In the scope detection, the results are also promising and  Accepted for publication in the Journal of the Association for Information Science and Technology. Pre-publication version. February 11, 2015. 2 represent a substantial improvement on the baseline (up by roughly 10%). A detailed error analysis is also provided. The extrinsic evaluation shows that the correct identification of cues and scopes is vital for the task of sentiment analysis.

The Visual Computer, 2010

This paper presents a novel concept: a graphical representation of human emotion extracted from text sentences. The major contributions of this paper are the following. First, we present a pipeline that extracts, processes, and renders emotion of 3D virtual human (VH). The extraction of emotion is based on data mining statistic of large cyberspace databases. Second, we propose methods to optimize this computational pipeline so that real-time virtual reality rendering can be achieved on common PCs. Third, we use the Poisson distribution to transfer database extracted lexical and language parameters into coherent intensities of valence and arousal-parameters of Russell's circumplex model of emotion. The last contribution is a practical color interpretation of emotion that influences the emotional aspect of rendered VHs. To test our method's efficiency, computational statistics related to classical or untypical cases of emotion are provided. In order to evaluate our approach, we applied our method to diverse areas such as cyberspace forums, comics, and theater dialogs.

TAL - Natural Language Processing, 2008

As any other text categorization task, authorship attribution requires a large number of training examples. These examples, which are easily ob- tained for most of the tasks, are particularly difficult to obtain for this case. Based on this fact, in this paper we investigate the possibility of using Web- based text mining methods for the identification of the author of

The world has widely changed in terms of communicating, acquiring, and storing information. Hundreds of millions of people are involved in information retrieval tasks on a daily basis, in particular while using a Web search engine or searching their e-mail, making such field the dominant form of information access, overtaking traditional database-style searching. How to handle this huge amount of information has now become a challenging issue. In this paper, after recalling the main topics concerning information retrieval, we present a survey on the main works on literature retrieval and mining in bioinformatics. While claiming that information retrieval approaches are useful in bioinformatics tasks, we discuss some challenges aimed at showing the effectiveness of these approaches applied therein.

Proc. Am. Soc. Info. Sci. Tech., 2011

Crowdsourcing has been recognized as a possible technique to complement costly user studies, usability studies, relevance judgment for information retrieval studies, and training set build-up for automatic document classification. However, the quality of crowdworkers varies by diverse factors and we often cannot tell whether their answers are right or wrong immediately due to the lack of gold standard answers. In this paper, we present a machine-learning based crowdworker filtering technique that can be used to assess workers immediately after they finish their assigned tasks. A Support Vector Machine (SVM)-based crowdworker filter, called a Smart Crowd Filter (SCFilter), was used to predict the probability that each label is correct and identifies those crowdworkers that consistently provide answers that are unlikely to be correct. To verify the performance of the SCFilter, a bad worker detection simulation test and an experiment in an actual crowdsourcing environment at the Amazon Mechanical Turk (AMT) website were performed. In the simulation test, bad worker detection performance was assessed in terms of precision and recall. In the experiment at the AMT website, a statistically significant improvement was observed for automatic document classification.

Journal of the American Society for Information Science and Technology, 2011

Sentiment analysis is a challenging new task related to text mining and natural language processing. Although there are, at present, several studies related to this theme, most of these focus mainly on English texts. The resources available for opinion mining (OM) in other languages are still limited. In this article, we present a new Arabic corpus for the OM task that has been made available to the scientific community for research purposes. The corpus contains 500 movie reviews collected from different web pages and blogs in Arabic, 250 of them considered as positive reviews, and the other 250 as negative opinions. Furthermore, different experiments have been carried out on this corpus, using machine learning algorithms such as support vector machines and Naïve Bayes. The results obtained are very promising and we are encouraged to continue this line of research.

Lecture Notes in Computer Science, 2009

In this paper we introduce and evaluate a technique for applying latent Dirichlet allocation to supervised semantic categorization of documents. In our setup, for every category an own collection of topics is assigned, and for a labeled training document only topics from its category are sampled. Thus, compared to the classical LDA that processes the entire corpus in one, we essentially build separate LDA models for each category with the category-specific topics, and then these topic collections are put together to form a unified LDA model. For an unseen document the inferred topic distribution gives an estimation how much the document fits into the category. We use this method for Web document classification. Our key results are 46% decrease in 1-AUC value in classification accuracy over tf.idf with SVM and 43% over the plain LDA baseline with SVM. Using a careful vocabulary selection method and a heuristic which handles the effect that similar topics may arise in distinct categories the improvement is 83% over tf.idf with SVM and 82% over LDA with SVM in 1-AUC. ⋆ Supported by the EU FP7 project LiWA -Living Web Archives and by grants OTKA NK 72845, ASTOR NKFP 2/004/05

Communications of the ACM, 2009

Pattern Analysis and Applications, 2015

Multi-label classification allows instances to belong to several classes at once. It has received significant attention in machine learning and has found many real world applications in recent years, such as text categorization, automatic video annotation and functional genomics, resulting in the development of many multi-label classification methods. Based on labelled examples in the training dataset, a multi-labelled method extracts inherent information in order to output a function that predicts the labels of unlabelled data. Due to several problems, like errors in the input vectors or in their labels, this information may be wrong and might lead the multilabel algorithm to fail. In this paper, we propose a simple algorithm for overcoming these problems by editing the existing training dataset, and adapting this edited set with different multi-label classification methods. Evaluation on benchmark datasets demonstrates the usefulness and effectiveness of our approach.

Journal of Information Science, 2014

Text Categorisation or Classification (TC) is concerned with placing text documents in their proper category according to their contents. Due to the various applications of TC and the large volume of text documents uploaded on the Internet daily, the need for such an automated method stems from the difficulty and tedium of doing such a process manually. The usefulness of TC is manifested in different fields and needs. For instance, the ability of automatically classifying an article or an email into its right class (Arts, Economics, Politics, Sports, etc.) would be very appreciated by individual users as well as companies. This paper is concerned with TC of Arabic articles. It contains a comparison of the five best known algorithms for TC. It also studies the effects of utilizing different Arabic stemmers (light and root-based stemmers) on the effectiveness of these classifiers. Furthermore, a comparison between different data mining software tools (Weka and RapidMiner) is presented. The results illustrate the strong accuracy provided by the SVM classifier, especially when used with the light10 stemmer. This outcome can be used in future as a baseline to compare with other unexplored classifiers and Arabic stemmers.

Natural Language Engineering, 2012

Vector space models are used in language processing applications for calculating semantic similarities of words or documents. The vector spaces are generated with feature extraction methods for text data. However, evaluation of the feature extraction methods may be difficult. Indirect evaluation in an application is often time-consuming and the results may not generalize to other applications, whereas direct evaluations that measure the amount of captured semantic information usually require human evaluators or annotated data sets. We propose a novel direct evaluation method based on canonical correlation analysis (CCA), the classical method for finding linear relationship between two data sets. In our setting, the two sets are parallel text documents in two languages. A good feature extraction method should provide representations that reflect the semantic content of the documents. Assuming that the underlying semantic content is independent of the language, we can study feature extraction methods that capture the content best by measuring dependence between the representations of a document and its translation. In the case of CCA, the applied measure of dependence is correlation. The evaluation method is based on unsupervised learning, it is languageand domain-independent, and it does not require additional resources besides a parallel corpus. In this paper, we demonstrate the evaluation method on a sentence-aligned parallel corpus. The method is validated by showing that the obtained results with bag-of-words representations are intuitive and agree well with the previous findings. Moreover, we examine * We are grateful to the anonymous reviewers for their detailed and insightful comments on this paper. We also thank our colleagues Marcus Dobrinkat, Timo Honkela, Arto Klami, Oskar Kohonen, and Jaakko Väyrynen for their feedback and advice. SV, MP, TL, and KL belong to the Adaptive

Empirical Software Engineering, 2014

Software repositories hold applications that are often categorized to improve the effectiveness of various maintenance tasks. Properly categorized applications allow stakeholders to identify requirements related to their applications and predict maintenance problems in software projects. Manual categorization is expensive, tedious, and laborious -this is why automatic categorization approaches are gaining widespread importance. Unfortunately, for different legal and organizational reasons, the applications' source code is often not available, thus making it difficult to automatically categorize these applications. In this paper, we propose a novel approach in which we use Application Programming Interface (API) calls from third-party libraries for automatic categorization of software applications that use these API calls. Our approach is general since it enables different categorization algorithms to be applied to repositories that contain both source code and bytecode of applications, since API calls can be extracted from both the source code and byte-code. We compare our approach to a state-of-the-art approach that uses machine learning algorithms for software categorization, and conduct experiments on two large Java repositories: an open-source repository containing 3,286 projects and a closed-source repository with 745 applications, where the source code was not available. Our contribution is twofold: we propose a new approach that makes it possible to categorize software projects without any source code using a small number of API calls as attributes, and furthermore we carried out a comprehensive empirical evaluation of automatic categorization approaches.

Lecture Notes in Computer Science, 2013

International Journal of Quantum Chemistry, 2015

Accurate approximations to density functionals have recently been obtained via machine learning (ML). By applying ML to a simple function of one variable without any random sampling, we extract the qualitative dependence of errors on hyperparameters. We find universal features of the behavior in extreme limits, including both very small and very large length scales, and the noise-free limit. We show how such features arise in ML models of density functionals.

Lecture Notes in Computer Science, 2005

We develop a novel user-centric modeling technology, which can dynamically describe and update a person's expertise profile. In an enterprise environment, the technology can enhance employees' collaboration and productivity by assisting in finding experts, training employees, etc. Instead of using the traditional search methods, such as the keyword match, we propose to use relational and evolutionary graph models, which we call ExpertiseNet, to describe and find experts. These ExpertiseNets are used for mining, retrieval, and visualization. We conduct experiments by building ExpertiseNets for researchers from a research paper collection. The experiments demonstrate that expertise mining and matching are more efficiently achieved based on the proposed relational and evolutionary graph models.

Journal of the American Medical Informatics Association, 2007

A b s t r a c t Cancer staging provides a basis for planning clinical management, but also allows for meaningful analysis of cancer outcomes and evaluation of cancer care services. Despite this, stage data in cancer registries is often incomplete, inaccurate, or simply not collected. This article describes a prototype software system (Cancer Stage Interpretation System, CSIS) that automatically extracts cancer staging information from medical reports. The system uses text classification techniques to train support vector machines (SVMs) to extract elements of stage listed in cancer staging guidelines. When processing new reports, CSIS identifies sentences relevant to the staging decision, and subsequently assigns the most likely stage. The system was developed using a database of staging data and pathology reports for 710 lung cancer patients, then validated in an independent set of 179 patients against pathologic stage assigned by two independent pathologists. CSIS achieved overall accuracy of 74% for tumor (T) staging and 87% for node (N) staging, and errors were observed to mirror disagreements between human experts.

Journal of Information Science, 2015

It was not until 2010 when businesses, politicians and people in general began to realise the potential of Twitter in Spain. This fact has awoken research interest in the extraction of knowledge from Twitter. This paper aims to fill the gap of the lack of resources for Twitter sentiment analysis in Spanish by performing a study of different features and machine learning algorithms for classifying the polarity of Twitter posts. The result is a new corpus of Spanish tweets called COST, and we have carried out a wide-ranging experimentation in which different machine learning algorithms have been used. Furthermore, we have tested the influence of using different weighting schemes for unigrams, the influence of eliminating stop-words and the application of a stemmer process.

Advances in Computer and Information Sciences and Engineering, 2008

Feature selection is essential for effective and accurate text classification systems. This paper investigates the effectiveness of six commonly used feature selection methods, Evaluation used an in-house collected Arabic text classification corpus, and classification is based on Support Vector Machine Classifier. The experimental results are presented in terms of precision, recall and Macroaveraged F 1 measure.

Handbook on Ontologies, 2009

To continually track changing skills and knowledge requirements for human factors professionals in the labor market, an archival human factors jobs database was created. The database contains relevant information from human factors job announcements, including required degree and field of education, required and preferred experience, and information from the job descriptions. The primary purpose of the database is to facilitate longitudinal analyses of employers’ expectations. Results from these analyses should be used in the design of academic human factors programs and curricula to better prepare future human factors professionals meet such expectations upon their graduation. In this paper we describe a natural language processing approach to analysis of human factors jobs data. Several trends emerged from our analyses. Although we defer drawing conclusions from these trends until later, until the database contains much more data that are more evenly distributed in time, our metho...

Studies in Computational Intelligence, 2010

The analysis of human behaviour has been conducted within diverse disciplines, such as psychology, sociology, economics, linguistics, marketing and computer science. Hence, a broad theoretical framework is available, with a high potential for application into other areas, in particular to the analysis of web user browsing behaviour. The above mentioned disciplines use surveys and experimental sampling for testing and calibrating their theoretical models. With respect to web user browsing behaviour, the major source of data is the web logs, which store every visitor's action on a web site. Such files could contain millions of registers, depending on the web site traffic, and represents a major data source about human behaviour. This chapter surveys the new trends in analysing web user behaviour and revises some novel approaches, such as those based on the neurophysiological theory of decision making, for describing what web users are looking for in a web site.

Journal of Business Finance & Accounting, 2011

This study extends the literature on the information content of stock message boards. To better understand the effect of online postings on trading activities and reduce the error due to stocks with small message board followings, we examine stocks with no fundamental news and high message posting activity. Such stocks tend to be of small firms with weak financials. For those stocks, we find a two-day pump followed by a two-day dump manipulation pattern among online traders, which suggests that an online stock message board can be used as a herding device to temporarily drive up stock prices. We also find that online traders' credit-weighted sentiment index, but not the number of postings, is positively associated with contemporaneous return and negatively predicts the return next day and two days later. Also, absolute sentiment is negatively related with contemporaneous and next day's intraday volatility and positively related with the proportion of volume in small-sized trades. We conclude that message board sentiment is an important predictor of trading-related activities.

The Structure of Style, 2010

As we have seen, there are many diverse concepts of what we mean when we speak of the 'style' of a work; we may speak, for example, of an individual's specific style of artistic expression , of style as a means of implicitly contextualizing a work as part of a certain genre , of style as a means of conveying emotion [26], or of style as a concept organizing how observers tend to interpret a work . In this chapter, we explore automated methods for stylistic analysis of written texts (computational stylistics), formulated as having roughly two research goals: (a) develop methods to automatically distinguish texts with a certain stylistic character from those of another, and (b) distill an interpretable representation of the difference between such stylistic characters. We seek an inclusive view of the nature of style, endeavoring to consider models and methods covering the individual style of a genius as well as the generic style of a collective, style's communicative functions as well as its social determinants, and the intentions of the author as well as the potential reaction of the reader.

Recommender Systems Handbook, 2010

Recommender systems have the effect of guiding users in a personalized way to interesting objects in a large space of possible options. Content-based recommendation systems try to recommend items similar to those a given user has liked in the past. Indeed, the basic process performed by a content-based recommender consists in matching up the attributes of a user profile in which preferences and interests are stored, with the attributes of a content object (item), in order to recommend to the user new interesting items. This chapter provides an overview of content-based recommender systems, with the aim of imposing a degree of order on the diversity of the different aspects involved in their design and implementation. The first part of the chapter presents the basic concepts and terminology of contentbased recommender systems, a high level architecture, and their main advantages and drawbacks. The second part of the chapter provides a review of the state of the art of systems adopted in several application domains, by thoroughly describing both classical and advanced techniques for representing items and user profiles. The most widely adopted techniques for learning user profiles are also presented. The last part of the chapter discusses trends and future research which might lead towards the next generation of systems, by describing the role of User Generated Content as a way for taking into account evolving vocabularies, and the challenge of feeding users with serendipitous recommendations, that is to say surprisingly interesting items that they might not have otherwise discovered.

Lecture Notes in Computer Science, 2009

Basic content-based personalization consists in matching up the attributes of a user profile, in which preferences and interests are stored, with the attributes of a content object. The Web 2.0 (r)evolution has changed the game for personalization, from 'elitary' Web 1.0, written by few and read by many, to web content generated by everyone (usergenerated content -UGC), since the role of people has evolved from passive consumers of information to that of active contributors. One of the forms of UGC that has drawn most attention of the research community is folksonomy, a taxonomy generated by users who collaboratively annotate and categorize resources of interests with freely chosen keywords called tags. FIRSt (Folksonomy-based Item Recommender syStem) is a contentbased recommender system developed at the University of Bari which integrates UGC (through social tagging) in a classic content-based model, letting users express their preferences for items by entering a numerical rating as well as to annotate rated items with free tags. FIRSt is capable of providing recommendations for items in several domains (e.g., movies, music, books), provided that descriptions of items are available as text documents (e.g. plot summaries, reviews, short abstracts). This paper describes the system general architecture and user modeling approach, showing how this recommendation model has been applied to recommend the artworks located at the Vatican Picture Gallery (Pinacoteca Vaticana), providing users with a personalized museum tour tailored on their tastes.

Public understanding of science (Bristol, England), 2015

This article presents our study of science coverage in the digital Spanish press over the last decade. We employed automated information retrieval procedures to create a corpus of 50,763 text units dealing with science and technology, and used automated text-analysis procedures in order to provide a general picture of the structure, characteristics and evolution of science news in Spain. We found between 6% and 7% of science coverage, a clear high proportion of biomedicine and predominance of science over technology, although we also detected an increase in technological content during the second half of the decade. Analysing the extrinsic and intrinsic features of science culture, we found a predominance of intrinsic features that still need further analysis. Our attempt to use specialised software to examine big data was effective, and allowed us to reach these preliminary conclusions.

Lecture Notes in Computer Science, 2014

Cylindrical algebraic decomposition(CAD) is a key tool in computational algebraic geometry, particularly for quantifier elimination over real-closed fields. When using CAD, there is often a choice for the ordering placed on the variables. This can be important, with some problems infeasible with one variable ordering but easy with another. Machine learning is the process of fitting a computer model to a complex function based on properties learned from measured data. In this paper we use machine learning (specifically a support vector machine) to select between heuristics for choosing a variable ordering, outperforming each of the separate heuristics.

Applied Artificial Intelligence, 2005

We describe a novel method for using Genetic Programming to create compact classification rules based on combinations of N-Grams (character strings). Genetic programs acquire fitness by producing rules that are effective classifiers in terms of precision and recall when evaluated against a set of training documents. We describe a set of functions and terminals and provide results from a classification task using the Reuters 21578 dataset. We also suggest that because the induced rules are meaningful to a human analyst they may have a number of other uses beyond classification and provide a basis for text mining applications.

Lecture Notes in Computer Science, 2006

This paper presents a content-based approach to spam detection based on low-level information. Instead of the traditional 'bag of words' representation, we use a 'bag of character n-grams' representation which avoids the sparse data problem that arises in n-grams on the word-level. Moreover, it is language-independent and does not require any lemmatizer or 'deep' text preprocessing. Based on experiments on Ling-Spam corpus we evaluate the proposed representation in combination with support vector machines. Both binary and term-frequency representations achieve high precision rates while maintaining recall on equally high level, which is a crucial factor for anti-spam filters, a cost sensitive application.

BMC Bioinformatics, 2008

Background: Synthesis of data from published human genetic association studies is a critical step in the translation of human genome discoveries into health applications. Although genetic association studies account for a substantial proportion of the abstracts in PubMed, identifying them with standard queries is not always accurate or efficient. Further automating the literaturescreening process can reduce the burden of a labor-intensive and time-consuming traditional literature search. The Support Vector Machine (SVM), a well-established machine learning technique, has been successful in classifying text, including biomedical literature. The GAPscreener, a free SVM-based software tool, can be used to assist in screening PubMed abstracts for human genetic association studies.

Topics in Cognitive Science, 2011

The focus of this paper is two-fold. First, similarities generated from six semantic models were compared to human ratings of paragraph similarity on two datasets -23 World Entertainment News Network paragraphs and 50 ABC newswire paragraphs. Contrary to findings on smaller textual units such as word associations (Grifths, Tenenbaum, & Steyvers, 2007), our results suggest that when single paragraphs are compared, simple non-reductive models (word overlap and vector space) can provide better similarity estimates than more complex models (LSA, Topic Model, SpNMF, and CSM). Second, various methods of corpus creation were explored to facilitate the semantic models' similarity estimates. Removing numeric and single characters, and also truncating document length improved performance. Automated construction of smaller Wikipedia-based corpora proved to be very effective even improving upon the performance of corpora that had been chosen for the domain. Model performance was further improved by augmenting corpora with dataset paragraphs.

Lecture Notes in Computer Science, 2010

In this paper we evaluate the performance of multilabel classification algorithms on the EUR-Lex database of legal documents of the European Union. On the same set of underlying documents, we defined three different large-scale multilabel problems with up to 4000 classes. On these datasets, we compared three algorithms: (i) the well-known one-against-all approach (OAA); (ii) the multiclass multilabel perceptron algorithm (MMP), which modifies the OAA ensemble by respecting dependencies between the base classifiers in the training protocol of the classifier ensemble; and (iii) the multilabel pairwise perceptron algorithm (MLPP), which unlike the previous algorithms trains one base classifier for each pair of classes. All algorithms use the simple but very efficient perceptron algorithm as the underlying classifier. This makes them very suitable for large-scale multilabel classification problems. While previous work has already shown that the latter approach outperforms the other two approaches in terms of predictive accuracy, its key problem is that it has to store one classifier for each pair of classes. The key contribution of this work is to demonstrate a novel technique that makes the pairwise approach feasible for problems with large number of classes, such as those studied in this work. Our results on the EUR-Lex database illustrate the effectiveness of the pairwise approach and the efficiency of the MMP algorithm. We also show that it is feasible to efficiently and effectively handle very large multilabel problems.

Advances in Intelligent Systems and Computing, 2014

The development of automatic methods to produce usable structured information from unstructured text sources is extremely valuable to the oil and gas industry. A structured resource would allow researches and industry professionals to write relatively simple queries to retrieve all the information regards transcriptions of any accident. Instead of the thousands of abstracts provided by querying the unstructured corpus, the queries on structured corpus would result in a few hundred well-formed results. On this paper we propose and evaluate information extraction techniques in occupational health control process, particularly, for the case of automatic detection of accidents from unstructured texts. Our proposal divides the problem in subtasks such as text analysis, recognition and classification of failed occupational health control, resolving accidents.

Lecture Notes in Computer Science, 2003

We propose mixtures of hidden Markov models for modelling clickstreams of web surfers. Hence, the page categorization is learned from the data without the need for a (possibly cumbersome) manual categorization. We provide an EM algorithm for training a mixture of HMMs and show that additional static user data can be incorporated easily to possibly enhance the labelling of users. Furthermore, we use prior knowledge to enhance generalization and avoid numerical problems. We use parameter tying to decrease the danger of overfitting and to reduce computational overhead. We put a flat prior on the parameters to deal with the problem that certain transitions between page categories occur very seldom or not at all, in order to ensure that a nonzero transition probability between these categories nonetheless remains. In applications to artificial data and real-world web logs we demonstrate the usefulness of our approach. We train a mixture of HMMs on artificial navigation patterns, and show that the correct model is being learned. Moreover, we show that the use of static 'satellite data' may enhance the labeling of shorter navigation patterns. When applying a mixture of HMMs to realworld web logs from a large Dutch commercial web site, we demonstrate that sensible page categorizations are being learned.

Lecture Notes in Computer Science, 2008

The performance of categorical music emotion classification that divides emotion into classes and uses audio features alone for emotion classification has reached a limit due to the presence of a semantic gap between the object feature level and the human cognitive level of emotion perception. Motivated by the fact that lyrics carry rich semantic information of a song, we propose a multi-modal approach to help improve categorical music emotion classification. By exploiting both the audio features and the lyrics of a song, the proposed approach improves the 4-class emotion classification accuracy from 46.6% to 57.1%. The results also show that the incorporation of lyrics significantly enhances the classification accuracy of valence.

Knowledge and Information Systems, 2011

The Feature Weighting Classifier (FWC) is an efficient multi-class classification algorithm for text data that uses Information Gain to directly estimate per-class feature weights in the classifier. This classifier requires only a single pass over the dataset to compute the feature frequencies per class, is easy to implement, and has memory usage that is linear in the number of features. Results of experiments performed on 128 binary and multi-class text and web datasets show that FWC's performance is at least comparable to, and often better than that of Naive Bayes, TWCNB, Winnow, Balanced Winnow and linear SVM. On a large-scale web dataset with 12,294 classes and 135,973 training instances, FWC trained in 13 seconds and yielded comparable classification performance to a state of the art multi-class SVM implementation, which took over 15 minutes to train.

Lecture Notes in Computer Science, 2005

The aim of the study was to determine how significance indicators assigned to different Web page elements (internal metadata, title, headings, and main text) influence automated classification. The data collection that was used comprised 1000 Web pages in engineering, to which Engineering Information classes had been manually assigned. The significance indicators were derived using several different methods: (total and partial) precision and recall, semantic distance and multiple regression. It was shown that for best results all the elements have to be included in the classification process. The exact way of combining the significance indicators turned out not to be overly important: using the F1 measure, the best combination of significance indicators yielded no more than 3% higher performance results than the baseline.

Lecture Notes in Computer Science, 2008

In this paper, we perform Chinese text classification using n-gram text representation on TanCorp which is a new large corpus special for Chinese text classification more than 14,000 texts divided into 12 classes. We use different n-gram feature (1-, 2-grams or 1-, 2-, 3-grams) to represent documents. Different feature weights (absolute text frequency, relative text frequency, absolute n-gram frequency and relative n-gram frequency) are compared. The sparseness of "document by feature" matrices is analyzed in various cases. We use the C-SVC classifier which is the SVM algorithm designed for the multi-classification task. We perform our experiments in the TANAGRA platform. We found out that the feature selection methods based on n-gram frequency (absolute or relative) always give better results and produce denser matrices.

International Journal on Artificial Intelligence Tools, 2006

Authorship attribution can assist the criminal investigation procedure as well as cybercrime analysis. This task can be viewed as a single-label multi-class text categorization problem. Given that the style of a text can be represented as mere word frequencies selected in a language-independent method, suitable machine learning techniques able to deal with high dimensional feature spaces and sparse data can be directly applied to solve this problem. This paper focuses on classifier ensembles based on feature set subspacing. It is shown that an effective ensemble can be constructed using, exhaustive disjoint subspacing, a simple method producing many poor but diverse base classifiers. The simple model can be enhanced by a variation of the technique of cross-validated committees applied to the feature set. Experiments on two benchmark text corpora demonstrate the effectiveness of the presented method improving previously reported results and compare it to support vector machines, an alternative suitable machine learning approach to authorship attribution.

Lecture Notes in Computer Science, 2006

Automatic authorship identification offers a valuable tool for supporting crime investigation and security. It can be seen as a multi-class, single-label text categorization task. Character n-grams are a very successful approach to represent text for stylistic purposes since they are able to capture nuances in lexical, syntactical, and structural level. So far, character n-grams of fixed length have been used for authorship identification. In this paper, we propose a variable-length n-gram approach inspired by previous work for selecting variable-length word sequences. Using a subset of the new Reuters corpus, consisting of texts on the same topic by 50 different authors, we show that the proposed approach is at least as effective as information gain for selecting the most significant n-grams although the feature sets produced by the two methods have few common members. Moreover, we explore the significance of digits for distinguishing between authors showing that an increase in performance can be achieved using simple text pre-processing.

Journal of Experimental & Theoretical Artificial Intelligence, 2007

An intelligent system should be able to solve a wide range of problems from different domains. In this paper we propose a complex and adaptive system capable of solving various data analysis problems without needing human help for parameter settings. The presented system, called A-Brain, consists of several inter-connected components (a Decision Maker, a Trainer and several Problem Solvers) which provide a base for building complex problem solvers. The parameters of the Trainer's algorithm are problem independent. This fact is a requirement for the intelligent systems that cannot count on the human intervention while operating. The A-Brain system is applied for solving some well-known problems in the field of symbolic regression and classification. Numerical experiments show that A-Brain system is able to perform very well on the considered test problems.

To educate the future human factors/ergonomics workforce and meet the demand for new professionals in the field, academic institutions must pay close attention to the ever-changing skills and knowledge expectations in the labor market. These trends are not easy to track, however. Surveys of new professionals about their experiences in their first jobs or surveys of employers about their experiences with new hires suffer from low response rates, nonresponse bias, and the one-time nature of survey research. A better way to track labor market trends is to continually analyze human factors job postings for education and experience requirements specified in them. This paper describes development of a database for that purpose. We also discuss ways of analyzing unstructured text data in the database. The results of analyses of these data include summary statistics of frequencies and their correlations, clusters of similar jobs, and a continually updated mathematical model to classify jobs...

Lecture Notes in Computer Science, 2007

Automated text categorisation systems learn a generalised hypothesis from large numbers of labelled examples. However, in many domains labelled data is scarce and expensive to obtain. Active learning is a technique that has shown to reduce the amount of training data required to produce an accurate hypothesis. This paper proposes a novel method of incorporating predictions made in previous iterations of active learning into the selection of informative unlabelled examples. We show empirically how this method can lead to increased classification accuracy compared to alternative techniques.

International Journal of Systems Science, 2006

Exploring digital collections to find information relevant to a user's interests is a challenging task. Algorithms designed to solve this relevant information problem base their relevance computations on user profiles in which representations of the users' interests are maintained. This article presents a new method, based on the classic Rocchio algorithm for text categorization, able to discover user preferences from the analysis of textual descriptions of items in online catalog of e-commerce Web sites. Experiments have been carried out on several data sets, and results have been compared with those obtained using an inductive logic programming (ILP) approach and a probabilistic one.

Journal of Intelligent Information Systems, 2013

The rapid growth of the so-called Web 2.0 has changed the surfers' behavior. A new democratic vision emerged, in which users can actively contribute to the evolution of the Web by producing new content or enriching the existing one with user generated metadata. In this context the use of tags, keywords freely chosen by users for describing and organizing resources, spread as a model for browsing and retrieving web contents. The success of that collaborative model is justified by two factors: firstly, information is organized in a way that closely reflects the users' mental model; secondly, the absence of a controlled vocabulary reduces the users' learning curve and allows the use of evolving vocabularies.

Journal of Documentation, 2006

Purpose -To provide an integrated perspective to similarities and differences between approaches to automated classification in different research communities (machine learning, information retrieval and library science), and point to problems with the approaches and automated classification as such. Design/methodology/approach -A range of works dealing with automated classification of full-text web documents are discussed. Explorations of individual approaches are given in the following sections: special features (description, differences, evaluation), application and characteristics of web pages. Findings -Provides major similarities and differences between the three approaches: document pre-processing and utilization of web-specific document characteristics is common to all the approaches; major differences are in applied algorithms, employment or not of the vector space model and of controlled vocabularies. Problems of automated classification are recognized.

Lecture Notes in Business Information Processing, 2007

This paper presents an algorithm to improve a web search query based on the feedback on the viewed documents. A user who is searching for information on the Web marks the retrieved (viewed) documents as relevant or irrelevant to further expose the information needs expressed in the original query. A new web search query matching this improved understanding of the user's information needs is synthesized from these text documents. The methodology provides a way for creating web search query that matches the user's information need even when the user may have difficulty in doing so directly due to lack of experience in the query design or lack of familiarity of the search domain. A user survey has shown that the algorithmically formed query has recall coverage and precision characteristics better than those achieved by the experienced human web searchers.

Machine Learning, 2012

We propose a general framework to incorporate first-order logic (FOL) clauses, that are thought of as an abstract and partial representation of the environment, into kernel machines that learn within a semi-supervised scheme. We rely on a multi-task learning scheme where each task is associated with a unary predicate defined on the feature space, while higher level abstract representations consist of FOL clauses made of those predicates. We re-use the kernel machine mathematical apparatus to solve the problem as primal optimization of a function composed of the loss on the supervised examples, the regularization term, and a penalty term deriving from forcing real-valued constraints deriving from the predicates. Unlike for classic kernel machines, however, depending on the logic clauses, the overall function to be optimized is not convex anymore. An important contribution is to show that while tackling the optimization by classic numerical schemes is likely to be hopeless, a stage-based learning scheme, in which we start learning the supervised examples until convergence is reached, and then continue by forcing the logic clauses is a viable direction to attack the problem. Some promising experimental results are given on artificial learning tasks and on the automatic tagging of bibtex entries to emphasize the comparison with plain kernel machines.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012

We demonstrate a metadata enrichment system for the Europeana digital library. The system allows different institutions which provide to Europeana pointers (in the form of metadata records -MRs) to their content to enrich their MRs by classifying them under a classification scheme of their choice, and to extract/highlight entities of significant interest within the MRs themselves. The use of a supervised learning metaphor allows each content provider (CP) to generate classifiers and extractors tailored to the CP's specific needs, thus allowing the tool to be effectively available to the multitude (2000+) of Europeana CPs.

Journal of Computational Biology, 2003

The past decade has seen a tremendous growth in the amount of experimental and computational biomedical data, specifically in the areas of genomics and proteomics. This growth is accompanied by an accelerated increase in the number of biomedical publications discussing the findings. In the last few years, there has been a lot of interest within the scientific community in literature-mining tools to help sort through this abundance of literature and find the nuggets of information most relevant and useful for specific analysis tasks. This paper provides a road map to the various literature-mining methods, both in general and within bioinformatics. It surveys the disciplines involved in unstructured-text analysis, categorizes current work in biomedical literature mining with respect to these disciplines, and provides examples of text analysis methods applied towards meeting some of the current challenges in bioinformatics.

Lecture Notes in Computer Science, 2007

In this paper, we propose a probabilistic approach to feature selection for multi-class text categorization. Specifically, we regard document class and occurrence of each feature as events, calculate the probability of occurrence of each feature by the theorem on the total probability and utilize the values as a ranking criterion. Experiments on Reuters-2000 collection show that the proposed method can yield better performance than information gain and χ-square, which are two wellknown feature selection methods.

Lecture Notes in Computer Science, 2011

We investigate the implementation of multi-label classification algorithms with a reject option, as a mean to reduce the time required to human annotators and to attain a higher classification accuracy on automatically classified samples than the one which can be obtained without a reject option. Based on a recently proposed model of manual annotation time, we identify two approaches to implement a reject option, related to the two main manual annotation methods: browsing and tagging. In this paper we focus on the approach suitable to tagging, which consists in withholding either all or none of the category assignments of a given sample. We develop classification reliability measures to decide whether rejecting or not a sample, aimed at maximising classification accuracy on non-rejected ones. We finally evaluate the trade-off between classification accuracy and rejection rate that can be attained by our method, on three benchmark data sets related to text categorisation and image annotation tasks.

Proceedings of the American Society for Information Science and Technology, 2013

Spoilers-critical plot information about works of fiction that "spoil" a viewer's enjoyment-have prompted elaborate conventions on social media to allow readers to insulate themselves from spoilers. However, these solutions depend on the conscientiousness and rigor of Internet posters and are thus an imperfect system. We create an automatic alternative that could alert users when a piece of text contains a spoiler. An automated spoiler detector serves not only as an additional protection against spoilers, but it also contributes to important problems in computational linguistics. We develop a new dataset of spoilers gathered from social media and create automatic classifiers using machine learning techniques. After establishing baseline performance using lexical features, we develop metadata-based features that substantially improve performance on the spoiler detection task.

EURASIP Journal on Advances in Signal Processing, 2011

Biometric discretization extracts a binary string from a set of real-valued features per user. This representative string can be used as a cryptographic key in many security applications upon error correction. Discretization performance should not degrade from the actual continuous features-based classification performance significantly. However, numerous discretization approaches based on ineffective encoding schemes have been put forward. Therefore, the correlation between such discretization and classification has never been made clear. In this article, we aim to bridge the gap between continuous and Hamming domains, and provide a revelation upon how discretization based on equal-width quantization and linearly separable subcode encoding could affect the classification performance in the Hamming domain. We further illustrate how such discretization can be applied in order to obtain a highly resembled classification performance under the general Lp distance and the inner product metrics. Finally, empirical studies conducted on two benchmark face datasets vindicate our analysis results.

Lecture Notes in Computer Science, 2006

This paper introduces a general methodology for performing distributed search in the Semantic Web. We propose to define this task as a three steps process, namely resource selection, query reformulation/ontology alignment and rank aggregation/data fusion. For the second problem, we have implemented oMAP, a formal framework for automatically aligning OWL ontologies. In oMAP, different components are combined for finding suitable mapping candidates (together with their weights), and the set of rules with maximum matching probability is selected. Among these components, traditional terminological-based classifiers, machine learning-based classifiers and a new classifier using the structure and the semantics of the OWL ontologies are proposed. oMAP has been evaluated on international test sets.

Online Information Review, 2009

Social tagging systems allow users to assign keywords (tags) to useful resources facilitating their future access by the tag creator, and possibly by other users. Social tagging has both proponents and critics, and this paper investigates if tags are an effective means for resource discovery.

International Journal of Web Information Systems, 2012

Computational Intelligence, 2012

Clustering Web data is one important technique for extracting knowledge from the Web. In this paper, a novel method is presented to facilitate the clustering. The method determines the appropriate number of clusters and provides suitable representatives for each cluster by inference from a Bayesian network. Furthermore, by means of the Bayesian network, the contents of the Web pages are converted into vectors of lower dimensions. The method is also extended for hierarchical clustering, and a useful heuristic is developed to select a good hierarchy. The experimental results show that the clusters produced benefit from high quality.

Knowledge and Information Systems, 2005

In this paper we propose a new algorithm for Feature Selection, called Best Terms (BT).

Pragmatics, 2011

Text mining aims at constructing classification models and finding interesting patterns in large text collections. This paper investigates the utility of applying these techniques to media analysis, more specifically to support discourse analysis of news reports about the 2007 Kenyan elections and postelection crisis in local (Kenyan) and Western (British and US) newspapers. It illustrates how text mining methods can assist discourse analysis by finding contrast patterns which provide evidence for ideological differences between local and international press coverage. Our experiments indicate that most significant differences pertain to the interpretive frame of the news events: whereas the newspapers from the UK and the US focus on ethnicity in their coverage, the Kenyan press concentrates on sociopolitical aspects.

Lecture Notes in Computer Science, 2008

This paper exposes the results of our participation in the Document Mining track at INEX'07. We have focused on the task of classification of XML documents. Our approach to deal with structured document representations uses classification methods for plain text, applied to flattened versions of the documents, where some of their structural properties have been translated to plain text. We have explored several options to convert structured documents into flat documents, in combination with two probabilistic methods for text categorization. The main conclusion of our experiments is that taking advantage of document structure to improve classification results is a difficult task.

Lecture Notes in Computer Science, 2007

We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). We apply the method to the classification of parliamentary initiatives in the regional Parliament of Andalucía at Spain from the Eurovoc thesaurus.

Computational and Mathematical Methods in Medicine, 2014

Guillain-Barré syndrome (GBS) is a neurological disorder which has not been explored using clustering algorithms. Clustering algorithms perform more efficiently when they work only with relevant features. In this work, we applied correlation-based feature selection (CFS), chi-squared, information gain, symmetrical uncertainty, and consistency filter methods to select the most relevant features from a 156-feature real dataset. This dataset contains clinical, serological, and nerve conduction tests data obtained from GBS patients. The most relevant feature subsets, determined with each filter method, were used to identify four subtypes of GBS present in the dataset. We used partitions around medoids (PAM) clustering algorithm to form four clusters, corresponding to the GBS subtypes. We applied the purity of each cluster as evaluation measure. After experimentation, symmetrical uncertainty and information gain determined a feature subset of seven variables. These variables conformed as...

Journal of biomedical semantics, 2018

Automatic identification of term variants or acceptable alternative free-text terms for gene and protein names from the millions of biomedical publications is a challenging task. Ontologies, such as the Cardiovascular Disease Ontology (CVDO), capture domain knowledge in a computational form and can provide context for gene/protein names as written in the literature. This study investigates: 1) if word embeddings from Deep Learning algorithms can provide a list of term variants for a given gene/protein of interest; and 2) if biological knowledge from the CVDO can improve such a list without modifying the word embeddings created. We have manually annotated 105 gene/protein names from 25 PubMed titles/abstracts and mapped them to 79 unique UniProtKB entries corresponding to gene and protein classes from the CVDO. Using more than 14 M PubMed articles (titles and available abstracts), word embeddings were generated with CBOW and Skip-gram. We setup two experiments for a synonym detection...

Lecture Notes in Computer Science, 2008

The use of domain ontologies is becoming increasingly popular in Medical Natural Language Processing Systems. A wide variety of knowledge bases in multiple languages has been integrated into the Unified Medical Language System (UMLS) to create a huge knowledge source that can be accessed with diverse lexical tools. MetaMap (and its java version MMTx) is a tool that allows extracting medical concepts from free text, but currently there not exists a Spanish version. Our ongoing research is centered on the application of biomedical concepts to cross-lingual text classification, what makes it necessary to have a Spanish MMTx available. We have combined automatic translation techniques with biomedical ontologies and the existing English MMTx to produce a Spanish version of MMTx. We have evaluated different approaches and applied several types of evaluation according to different concept representations for text classification. Our results prove that the use of existing translation tools such as Google Translate produce translations with a high similarity to original texts in terms of extracted concepts.

Mathematics in Computer Science

Cylindrical Algebraic Decomposition (CAD) is a key tool in computational algebraic geometry, best known as a procedure to enable Quantifier Elimination over real-closed fields. However, it has a worst case complexity doubly exponential in the size of the input, which is often encountered in practice. It has been observed that for many problems a change in algorithm settings or problem formulation can cause huge differences in runtime costs, changing problem instances from intractable to easy. A number of heuristics have been developed to help with such choices, but the complicated nature of the geometric relationships involved means these are imperfect and can sometimes make poor choices. We investigate the use of machine learning (specifically support vector machines) to make such choices instead. Machine learning is the process of fitting a computer model to a complex function based on properties learned from measured data. In this paper we apply it in two case studies: the first to select between heuristics for choosing a CAD variable ordering; the second to identify when a CAD problem instance would benefit from Gröbner Basis preconditioning. These appear to be the first such applications of machine learning to Symbolic Computation. We demonstrate in both cases that the machine learned choice outperforms human developed heuristics.

Advances in Web Technologies and Engineering

The wide use of ontology in different applications has resulted in a plethora of automatic approaches for population and enrichment of an ontology. Ontology enrichment is an iterative process where the existing ontology is continuously updated with new concepts. A key aspect in ontology enrichment process is the concept learning approach. A learning approach can be a linguistic-based, statistical-based, or hybrid-based that employs both linguistic as well as statistical-based learning approaches. This chapter presents a concept enrichment model that combines contextual and semantic information of terms. The proposed model called SEMCON employs a hybrid concept learning approach utilizing functionalities from statistical and linguistic ontology learning techniques. The model introduced for the first time two statistical features that have shown to improve the overall score ranking of highly relevant terms for concept enrichment. The chapter also gives some recommendations and possible future research directions based on the discussion in following sections.

Frontiers in Digital Humanities

This article presents the results of a multidisciplinary project aimed at better understanding the impact of different digitization strategies in computational text analysis. More specifically, it describes an effort to automatically discern the authorship of Jacob and Wilhelm Grimm in a body of uncorrected correspondence processed by HTR (Handwritten Text Recognition) and OCR (Optical Character Recognition), reporting on the effect this noise has on the analyses necessary to computationally identify the different writing style of the two brothers. In summary, our findings show that OCR digitization serves as a reliable proxy for the more painstaking process of manual digitization, at least when it comes to authorship attribution. Our results suggest that attribution is viable even when using training and test sets from different digitization pipelines. With regards to HTR, this research demonstrates that even though automated transcription significantly increases the risk of text misclassification when compared to OCR, a cleanliness above ≈ 20% is already sufficient to achieve a higher-than-chance probability of correct binary attribution.

Chaos: An Interdisciplinary Journal of Nonlinear Science

Empirical data on real complex systems are becoming increasingly available. Parallel to this is the need for new methods of reconstructing (inferring) the structure of networks from time-resolved observations of their node-dynamics. The methods based on physical insights often rely on strong assumptions about the properties and dynamics of the scrutinized network. Here, we use the insights from machine learning to design a new method of network reconstruction that essentially makes no such assumptions. Speci cally, we interpret the available trajectories (data) as "features" and use two independent feature ranking approaches-Random Forest and RReliefF-to rank the importance of each node for predicting the value of each other node, which yields the reconstructed adjacency matrix. We show that our method is fairly robust to coupling strength, system size, trajectory length, and noise. We also nd that the reconstruction quality strongly depends on the dynamical regime.

Journalism Studies

Using a new analytical tool, supervised machine learning (SML), a large number of newspaper articles is analysed to answer the question how newspapers frame the news of public risks, in this case of earthquakes caused by gas drilling in The Netherlands. SML enabled the study of 2265 news articles published over a period of 25 years. Our study shows that there is a disproportional relation between media reporting and actual risk; and that the use of dramatization bias in framing the issue about gas drilling increased, but the use of personalization and negativity bias did not become more dominant after a major media change in 2013. Sensational/tabloid newspapers make more use of personalization bias, whereas quality newspapers make more use of value conflict and political disagreement in the framing about gas drilling.

Quality & Quantity

This study aims to provide a more sensitive understanding of the dynamics and tipping points of issue attention in news media by combining the strengths of quantitative and qualitative research. The topic of this 25-year longitudinal study is the volume and the content of newspaper articles about the emerging risk of gas drilling in The Netherlands. We applied supervised machine learning (SML) because this allowed us to study changes in the quantitative use of subtopics at the detailed sentence level in a large number of articles. The study shows that the actual risk of drilling-induced seismicity gradually increased and that the volume of newspaper attention for the issue also gradually increased for two decades. The sub-topics extracted from media articles during the low media attention period, covering factual information, can be interpreted as a part of episodic frame patterns about the drilling and its consequences. However, a sudden major shift in newspaper attention can be ob...

Journal of Computers in Education, 2015

A growing body of empirical evidence suggests that the adaptive capabilities of computer-based learning environments can be improved through the use of educational data mining techniques. Log-file trace data provides a wealth of information about learner behaviors that can be captured, monitored, and mined for the purposes of discovering new knowledge and detecting patterns of interest. This study aims to leverage these analytical techniques to mine learner behaviors in relation to both diagnostic reasoning processes and outcomes in BioWorld, a computer-based learning environment that support learners to practice solving medical problems and receive formative feedback. In doing so, hidden Markov models are used to model behavioral indicators of proficiency during problem solving, while an ensemble of text classification algorithms are applied to written case summaries that learners' write as an outcome of solving a case in BioWorld. The application of these algorithms characterize learner behaviors at different phases of problem solving which provides corroborating evidence in support of where revisions can be made to provide design guidelines of the system. We conclude by discussing the instructional design and pedagogical implications for the

ACM Transactions on the Web, 2009

We propose a methodology for building a robust query classification system that can identify thousands of query classes, while dealing in real time with the query volume of a commercial Web search engine. We use a pseudo relevance feedback technique: given a query, we determine its topic by classifying the Web search results retrieved by the query. Motivated by the needs of search advertising, we primarily focus on rare queries, which are the hardest from the point of view of machine learning, yet in aggregate account for a considerable fraction of search engine traffic. Empirical evaluation confirms that our methodology yields a considerably higher classification accuracy than previously reported. We believe that the proposed methodology will lead to better matching of online ads to rare queries and overall to a better user experience.

Computers

Processing of social media text like tweets is challenging for traditional Natural Language Processing (NLP) tools developed for well-edited text due to the noisy nature of such text. However, demand for tools and resources to correctly process such noisy text has increased in recent years due to the usefulness of such text in various applications. Literature reports various efforts made to develop tools and resources to process such noisy text for various languages, notably, part-of-speech (POS) tagging, an NLP task having a direct effect on the performance of other successive text processing activities. Still, no such attempt has been made to develop a POS tagger for Urdu social media content. Thus, the focus of this paper is on POS tagging of Urdu tweets. We introduce a new tagset for POS-tagging of Urdu tweets along with the POS-tagged Urdu tweets corpus. We also investigated bootstrapping as a potential solution for overcoming the shortage of manually annotated data and present...

Social Computing and Social Media. Applications and Analytics

Social media platforms in the last few years have facilitated the development of communities that discuss real-world events, and have shaped the way users interact. The content generated in these platforms reflect a variety of intentions, ranging from social interaction to commercial interest, among many others. The present study aims at the implementation of an automatic intent classification system for a Chilean electricity company social media account. The dataset was created from 5000 tweets that were manually classified by 5 people. If discrepancies were detected, a majority voting scheme was used in order to tag the tweets' intentions. In order to perform the experimental validation of the automatic classification with the machine learning algorithms, several text representations were used (tf-idf, tf-rfl and bin-rfl). The results obtained from the various tests that were conducted yielded satisfactory results. We also analyzed how to assign automatic responses to frequently asked questions, and obtained promising results.

With the ever increasing of internet applications and social networking sites, people nowadays can easily express their feelings towards any products and services. These online reviews act as an important source for further analysis and improved decision making. These reviews are mostly unstructured by nature and thus, need processing like sentiment analysis and classification to provide a meaningful information for future uses. In text analysis tasks, the appropriate selection of words/features will have a huge impact on the effectiveness of the classifier. Thus, this paper explores the effect of the pre-processing strategies in the sentiment analysis of online movie reviews. In this paper, supervised machine learning method was used to classify the reviews. The support vector machine (SVM) with linear and non-linear kernel has been considered as classifier for the classification of the reviews. The performance of the classifier is critically examined based on the results of precision, recall, f-measure, and accuracy. Two different features representations were used which are term frequency and term frequency-inverse document frequency. Results show that the pre-processing strategies give a significant impact on the classification process.

Mathematical and Computational Applications

Growth in the area of opinion mining and sentiment analysis has been rapid and aims to explore the opinions or text present on different platforms of social media through machine-learning techniques with sentiment, subjectivity analysis or polarity calculations. Despite the use of various machine-learning techniques and tools for sentiment analysis during elections, there is a dire need for a state-of-the-art approach. To deal with these challenges, the contribution of this paper includes the adoption of a hybrid approach that involves a sentiment analyzer that includes machine learning. Moreover, this paper also provides a comparison of techniques of sentiment analysis in the analysis of political views by applying supervised machine-learning algorithms such as Naïve Bayes and support vector machines (SVM).

Applied Sciences

Breast cancer is one of the most prevalent cancer types with a high mortality rate in women worldwide. This devastating cancer still represents a worldwide public health concern in terms of high morbidity and mortality rates. The diagnosis of breast abnormalities is challenging due to different types of tissues and textural variations in intensity. Hence, developing an accurate computer-aided system (CAD) is very important to distinguish normal from abnormal tissues and define the abnormal tissues as benign or malignant. The present study aims to enhance the accuracy of CAD systems and to reduce its computational complexity. This paper proposes a method for extracting a set of statistical features based on curvelet and wavelet sub-bands. Then the binary grey wolf optimizer (BGWO) is used as a feature selection technique aiming to choose the best set of features giving high performance. Using public dataset, Digital Database for Screening Mammography (DDSM), different experiments hav...

Journal of Physics: Conference Series

The kidneys are one of the most important organs including the excretion system in humans. The kidneys are responsible for maintaining blood concentrations to remain constant (homeostatic) and help to control blood pressure (BP). If the task of the kidney is not functioning properly it will cause kidney failure. In the past decade, data mining methods have been used to diagnose kidney failure. The dataset used to predict kidney failure was successfully summarized by Soundarapandian, and was named the Chronic Kidney Disease (CKD) dataset. But the data in the CKD dataset contains missing value and high dimension data (original data) so that it affects the evaluation results on classification. This research proposes methods in preprocessing data, namely modus in every class (MEC) method to solve missing value problems, and the weight information gain (WIG) method for solving high dimensional data problems, the proposed method is named the MEC + WIG method. The MEC + WIG method will be compared with the original method and the MEC method and evaluated based on the accuracy of the traditional classification method (k-NN, Naïve Bayes, C4.5, and CART). The results showed that the average accuracy of the MEC + WIG method was better than the original method and the MEC method, with the average accuracy of the MEC + WIG method at 98.13%, while the average value of the accuracy of the original method and MEC respectively amounting to 88.56% and 92.88%. There were significant differences between the three methods when tested using Friedman test with a p-value of 0.02. It can be concluded that the MEC + WIG method can improve the performance of traditional methods k-NN, Naive Bayes, C4.5 and CART by overcoming the problem of missing value and data high dimension.

International Journal of Software Innovation

The knowledge extraction process from biological data is increasingly being considered, it addresses general issues such as grouping, classification and association; The Protein classification is an important activity for the biologist to respond to biological needs. For this reason, the authors present a global framework inspired by the knowledge extraction process from biological data to classified proteins from their primary structure based on the association rules. This framework has three main steps: The first one is, the pre-processing phase, consists of extracting descriptors by N-Gram technique. The second is the extraction of associations rules, applying the Apriori algorithm. The third step is selecting the relevant rules, and applied the classifier. The experiments of this technique were performed on five classes of protein, extracted from UniProt data bank and compared with five classification methods in the WEKA platform. The obtained results satisfied the authors' ...

Journal of Agricultural and Environmental Ethics

Public's concern about poor animal welfare provided by intensive farming systems has increased over the last decades. This study reviewed the interest of the scientific research on the pain issue in pig production to assess if the societal instances may be a driving force for the research activity. A literature search protocol was set up to identify the peer-reviewed papers published between 1970 and 2017 that covered the topic of 'pain in pigs' using Scopus ® , database of Elsevier©. One hundred and thirty papers were selected and they were mainly focused on the practice of castration (64%) followed by tail docking (24%). The scientific community first focused on these painful practices as a way to improve production efficiency and quality issues while more recently, due to the increased pressure by the public opinion, turned its interest towards the search of alternative solutions. A text mining analysis on the abstract of the selected papers clearly indicated the effort of the research to explore solutions to alleviate pain. Evocative words of this target were the selected terms 'pharmacological analgesic' and 'anaesthetic treatments'. The text mining highlighted vocalizations as the main pain indicators in pigs as this term was frequently associated to 'acute stress'. Ethical issues were a minor research topic in the scientific literature on pig breeding but in the short run, they are supposed to become a major subject to justify the acceptance of the modern production systems at the eyes of the consumers.

Bilgi Yönetimi, 2019

Technological opportunities and an increase in the production of records have made inevitable the necessity of new management methods. Documents produced in public institutions in Turkey are organized and managed according to the Standard File Plan. Under the relevant law, it is mandatory to determine the subject of official correspondence from the File Plan and add the relevant codes to the records. The correct selection of these codes is essential for the healthy operating of the research-investigation processes and the successful completion of the access processes. However, incorrect codes have been given depending on institutional, personal, or managerial conditions that will interrupt the life cycle of records. Artificial intelligence applications can be utilized to minimize such misapplications and to make records classification more powerful.

Applied Sciences

With the growth of online information and sudden expansion in the number of electronic documents provided on websites and in electronic libraries, there is difficulty in categorizing text documents. Therefore, a rule-based approach is a solution to this problem; the purpose of this study is to classify documents by using a rule-based. This paper deals with the rule-based approach with the embedding technique for a document to vector (doc2vec) files. An experiment was performed on two data sets Reuters-21578 and the 20 Newsgroups to classify the top ten categories of these data sets by using a document to vector rule-based (D2vecRule). Finally, this method provided us a good classification result according to the F-measures and implementation time metrics. In conclusion, it was observed that our algorithm document to vector rule-based (D2vecRule) was good when compared with other algorithms such as JRip, One R, and ZeroR applied to the same Reuters-21578 dataset.

Studies in Fuzziness and Soft Computing, 2005

Making correct decisions often requires analysing large volumes of textual information. Text Mining is a budding new field that endeavours to garner meaningful information from natural language text. Text Mining is the process of applying automatic methods to analyse and structure textual data in order to create useable knowledge from previously unstructured information. Text Mining is inherently interdisciplinary, borrowing heavily from neighbouring fields such as data mining and computational linguistics. Some real application to define the state-of-the-art in Text Mining and to single out future needs and scenarios are collected. 17 cases 12 cases 12 cases 13 cases 16 cases 13 cases 17 cases 100 View publication stats View publication stats

Organizational Research Methods

Organizations are increasingly interested in classifying texts or parts thereof into categories, as this enables more effective use of their information. Manual procedures for text classification work well for up to a few hundred documents. However, when the number of documents is larger, manual procedures become laborious, time-consuming, and potentially unreliable. Techniques from text mining facilitate the automatic assignment of text strings to categories, making classification expedient, fast, and reliable, which creates potential for its application in organizational research. The purpose of this article is to familiarize organizational researchers with text mining techniques from machine learning and statistics. We describe the text classification process in several roughly sequential steps, namely training data preparation, preprocessing, transformation, application of classification techniques, and validation, and provide concrete recommendations at each step. To help resea...

ACM Transactions on Cyber-Physical Systems

The rapid increase in the number of malicious programs has made malware forensics a daunting task and caused users’ systems to become in danger. Timely identification of malware characteristics including its origin and the malware sample family would significantly limit the potential damage of malware. This is a more profound risk in Cyber-Physical Systems (CPSs), where a malware attack may cause significant physical damage to the infrastructure. Due to limited on-device available memory and processing power in CPS devices, most of the efforts for protecting CPS networks are focused on the edge layer, where the majority of security mechanisms are deployed. Since the majority of advanced and sophisticated malware programs are combining features from different families, these malicious programs are not similar enough to any existing malware family and easily evade binary classifier detection. Therefore, in this article, we propose a novel multilabel fuzzy clustering system for malware...

BioMed Research International, 2016

According to the Ministry of Labor, Health and Social Affairs of Georgia a new health management system has to be introduced in the nearest future. In this context arises the problem of structuring and classifying documents containing all the history of medical services provided. The present work introduces the instrument for classification of medical records based on the Georgian language. It is the first attempt of such classification of the Georgian language based medical records. On the whole 24.855 examination records have been studied. The documents were classified into three main groups (ultrasonography, endoscopy, and X-ray) and 13 subgroups using two well-known methods: Support Vector Machine (SVM) andK-Nearest Neighbor (KNN). The results obtained demonstrated that both machine learning methods performed successfully, with a little supremacy of SVM. In the process of classification a “shrink” method, based on features selection, was introduced and applied. At the first stag...

Lecture Notes in Computer Science, 2005

Exploring digital collections to find information relevant to a user's interests is a challenging task. Algorithms designed to solve this relevant information problem base their relevance computations on user profiles in which representations of the users' interests are maintained. This paper presents a new method, based on the classical Rocchio algorithm for text categorization, able to discover user preferences from the analysis of textual descriptions of items in online catalogues of ecommerce Web sites. Experiments have been carried out on a dataset of real users, and results have been compared with those obtained using an Inductive Logic Programming (ILP) approach and a probabilistic one.

Journal of healthcare engineering, 2018

More than 1 billion people suffer from chronic respiratory diseases worldwide, accounting for more than 4 million deaths annually. Inhaled corticosteroid is a popular medication for treating chronic respiratory diseases. Its side effects include decreased bone mineral density and osteoporosis. The aims of this study are to investigate the association of inhaled corticosteroids and fracture and to design a clinical support system for fracture prediction. The data of patients aged 20 years and older, who had visited healthcare centers and been prescribed with inhaled corticosteroids within 2002-2010, were retrieved from the National Health Insurance Research Database (NHIRD). After excluding patients diagnosed with hip fracture or vertebrate fractures before using inhaled corticosteroid, a total of 11645 patients receiving inhaled corticosteroid therapy were included for this study. Among them, 1134 (9.7%) were diagnosed with hip fracture or vertebrate fracture. The statistical result...

ACM Computing Surveys, 2015

Multilabel learning has become a relevant learning paradigm in the past years due to the increasing number of fields where it can be applied and also to the emerging number of techniques that are being developed. This article presents an up-to-date tutorial about multilabel learning that introduces the paradigm and describes the main contributions developed. Evaluation measures, fields of application, trending topics, and resources are also presented.

Applied Sciences

Contemporary education is a vast field that is concerned with the performance of education systems. In a formal e-learning context, student dropout is considered one of the main problems and has received much attention from the learning analytics research community, which has reported several approaches to the development of models for the early prediction of at-risk students. However, maximizing the results obtained by predictions is a considerable challenge. In this work, we developed a solution using only students’ interactions with the virtual learning environment and its derivative features for early predict at-risk students in a Brazilian distance technical high school course that is 103 weeks in duration. To maximize results, we developed an elitist genetic algorithm based on Darwin’s theory of natural selection for hyperparameter tuning. With the application of the proposed technique, we predicted the student at risk with an Area Under the Receiver Operating Characteristic C...

Journal of Internet Services and Applications

There are several educational resources distributed in different repositories that address to a wide range of subjects and different educational goals. The proper choice of these educational resources is a challenge. Recommendation systems may help users in this task. In order to generate personalized recommendations, it is important to identify information that will help to define user profile and assist in identifying his/her interests. The constant and ever-increasing use of social networks allows the identification of different information about profile, interests, preferences, style and behavior from the spontaneous interaction. This paper presents an infrastructure able to extract users' profile and educational context, from the Facebook social network and recommend educational resources. The proposal is supported by Information Extraction Techniques and Semantic Web technologies for extraction, enrichment and definition of user's profile and interests. The recommendation approach is based on learning objects repositories, linked data and video repositories. It takes advantage of user's spent time at the web. The proposal evaluation was made from the development of a prototype, three proofs of concept and a case study. The evaluation showed users' acceptance of extracted information about their educational interests, automatically generated from social network and enriched to find implicit interests. It was also validated the possibility of people recommendation, enabling the establishment of interest network, based on a specific subject, showing good partners to study and research.

Research in Social Change

The aim of this paper is the analysis and a presentatition of texts and terminological ontology of scientific and professional publications in the field of regional Industrial Symbiosis and Industrial Symbiotic Networks. With computer-aided text analysis it is possible to have a comprehensive overview of the publications of works in this field and a more detailed presentation of the most fruitful concepts of publications.

Lecture Notes in Computer Science, 2007

The amount of information available online is increasing exponentially. While this information is a valuable resource, its sheer volume limits its value. Many research projects and companies are exploring the use of personalized applications that manage this deluge by tailoring the information presented to individual users. These applications all need to gather, and exploit, some information about individuals in order to be effective. This area is broadly called user profiling. This chapter surveys some of the most popular techniques for collecting information about users, representing, and building user profiles. In particular, explicit information techniques are contrasted with implicitly collected user information using browser caches, proxy servers, browser agents, desktop agents, and search logs. We discuss in detail user profiles represented as weighted keywords, semantic networks, and weighted concepts. We review how each of these profiles is constructed and give examples of projects that employ each of these techniques. Finally, a brief discussion of the importance of privacy protection in profiling is presented. This chapter discusses user profiles specifically designed for providing personalized information access. Other types of profiles, build using different construction techniques, are described elsewhere in this book. In particular, Chapter 4 [40] discusses generic user modeling systems that are broader in scope, not necessarily focused on Internet applications. Related research on collaborative recommender systems, discussed in Chapter 9 of this book [81], combines information from multiple users in order to provide improved information services. Concern over privacy protection is growing in parallel with the demand for personalized features. These two trends seem to be in direct opposition to each other, so privacy protection must be a crucial component of every personalization system. A detailed discussion can be found in Chapter 21 of this book [39]. There are a wide variety of applications to which personalization can be applied and a wide variety of different devices available on which to deliver the personalized information. Early personalization research focused on personalized filtering and/or rating systems for e-mail [49], electronic newspapers [14, 16], Usenet newsgroups [41, 58, 86, 91, 106], and Web documents [4]. More recently, personalization efforts have focused on improving navigation effectiveness by providing browsing assistants [9, 13], and adaptive Web sites [69]. Because search is one of the most common activities performed today, many projects are now focusing on personalized Web search [46, 88, 92] and more details on the subject can be found in Chapter 6 of this book [52]. However, personalized approaches to searching other types of collections, e.g., short stories [76], Java source code [100], and images [14] have also been explored. Commercial products are also adopting personalized features, for example, Yahoo!'s personalized Web portals [110] and Google Lab's personalized search [30]. The aforementioned systems are just a few examples that illustrate the breadth of applications to which personalized approaches are being investigated. Nichols [63] and Oard and Marchionini [64] provide a general overview of some the issues and approaches to personalized rating and filtering and Pretschner [71] describes approximately 45 personalization systems. Most personalization systems are based on some type of user profile, a data instance of a user model that is applied to adaptive interactive systems. User profiles may include demographic information, e.g., name, age, country, education level, etc, and may also represent the interests or preferences of either a group of users or a single person. Personalization of Web portals, for example, may focus on individual users, for example, displaying news about specifically chosen topics or the market summary of specifically selected stocks, or a groups of users for whom distinctive characteristics where identified, for example, displaying targeted advertising on ecommerce sites. In order to construct an individual user's profile, information may be collected explicitly, through direct user intervention, or implicitly, through agents that monitor user activity. Although profiles are typically built only from topics of interest to the user, some projects have explored including information about non-relevant topics in the profile [35, 104]. In these approaches, the system is able to use both kinds of topics to identify relevant documents and discard non-relevant documents at the same time. Profiles that can be modified or augmented are considered dynamic, in contrast to static profiles that maintain the same information over time. Dynamic profiles that Explicit info Data Collection Technology Or Application Profile Constructor User Implicit info Keyword profile Semantic Net profile Concept profile Personalized Services As shown in Figure 2.1, the user profiling process generally consists of three main phases. First, an information collection process is used to gather raw information about the user. As described in Section 2.2, depending on the information collection process selected, different types of user data can be extracted. The second phase focuses on user profile construction from the user data. Section 2.3 summarizes a variety of ways in which profiles may be represented and Section 2.4 some of the ways a profile may be constructed. The final phase, in which a technology or application exploits information in the user profile in order to provide personalized services, is discussed in Parts II and III of this book. 2.2 Collecting Information About Users The first phase of a profiling technique collects information about individual users. A basic requirement of such a system is that it must be able to uniquely identify users. This task is described in more detail in Section 2.2.1. The information collected may be explicitly input by the user or implicitly gathered by a software agent. It may be collected on the user's client machine or gathered by the application server itself. Depending on how the information is collected, different data about the users may be extracted. Several options, and their impacts, are discussed in Section 2.2.2. In

Scientific Programming

Information is exploding on the web at exponential pace, so online movie review is becoming a substantial information resource for online users. However, users post millions of movie reviews on regular basis, and it is not possible for users to summarize the reviews. Movie review classification and summarization is one of the challenging tasks in natural language processing. Therefore, an automatic approach is demanded to summarize the vast amount of movie reviews, and it will allow the users to speedily distinguish the positive and negative aspects of a movie. This study has proposed an approach for movie review classification and summarization. For movie review classification, bag-of-words feature extraction technique is used to extract unigrams, bigrams, and trigrams as a feature set from given review documents, and represent the review documents as a vector space model. Next, the Naïve Bayes algorithm is employed to classify the movie reviews (represented as a feature vector) in...

Lecture Notes in Computer Science, 2015

We perform an empirical study to explore the role of evolutionary linguistics on the text classification problem. We conduct experiments on a real-world collection with more than 100.000 Dutch historical notary acts. The document collection spans over six centuries. During such a large time period some lexical terms modified significantly. Person names, professions and other information changed over time as well. Standard text classification techniques which ignore temporal information of the documents might not produce the most optimal results in our case. Therefore, we analyse the temporal aspects of the corpus. We explore the effect of training and testing the model on different time periods. We use time periods that correspond to the main historical events and also apply clustering techniques in order to create time periods in a data driven way. All experiments show a strong time-dependency of our corpus. Exploiting this dependence, we extend standard classification techniques by combining different models trained on particular time periods and achieve overall accuracy above 90% and macro-average indicators above 63%.

Communications in Computer and Information Science, 2016

This paper presents an approach for automatically retrieving family relationships from a real-world collection of Dutch historical notary acts. We aim to retrieve relationships like husband-wife, parent-child, widow of, etc. Our approach includes person names extraction, reference disambiguation, candidate generation and family relationship prediction. Since we have a limited amount of training data, we evaluate different feature configurations based on the n-gram analysis. The best results were obtained by using a combination of bi-grams and trigrams of words together with the distance in words between two names. We evaluate our results for each type of the relationships in terms of precision, recall and f − score.

Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval

We transfer a key idea from the field of sentiment analysis to a new domain: community question answering (cQA). e cQA task we are interested in is the following: given a question and a thread of comments, we want to re-rank the comments, so that the ones that are good answers to the question would be ranked higher than the bad ones. We notice that good vs. bad comments use specific vocabulary and that one can o en predict the goodness/badness of a comment even ignoring the question, based on the comment contents only. is leads us to the idea to build a good/bad polarity lexicon as an analogy to the positive/negative sentiment polarity lexicons, commonly used in sentiment analysis. In particular, we use pointwise mutual information in order to build large-scale goodness polarity lexicons in a semi-supervised manner starting with a small number of initial seeds. e evaluation results show an improvement of 0.7 MAP points absolute over a very strong baseline, and state-of-the art performance on SemEval-2016 Task 3. CCS CONCEPTS •Computing methodologies → Natural language processing; •Information systems → estion answering; Retrieval models and ranking; Learning to rank;

Proceedings of the 15th International Conference on Mining Software Repositories

A recent research trend has emerged to identify developers' emotions, by applying sentiment analysis to the content of communication traces left in collaborative development environments. Trying to overcome the limitations posed by using off-the-shelf sentiment analysis tools, researchers recently started to develop their own tools for the software engineering domain. In this paper, we report a benchmark study to assess the performance and reliability of three sentiment analysis tools specifically customized for software engineering. Furthermore, we offer a reflection on the open challenges, as they emerge from a qualitative analysis of misclassified texts. 1 CCS CONCEPTS • Collaborative and social Computing → Collaborative and social computing theory, concepts and paradigms; Computer supported cooperative work; • Information retrieval → Retrieval tasks and goals; Sentiment analysis; • Software creation and management → Collaboration in software development; Programming teams

RadioGraphics

Natural language processing (NLP) is a methodology designed to extract concepts and meaning from human-generated unstructured (free-form) text. It is intended to be implemented by using computer algorithms so that it can be run on a corpus of documents quickly and reliably. To enable machine learning (ML) techniques in NLP, free-form text must be converted to a numerical representation. After several stages of preprocessing including tokenization, removal of stop words, token normalization, and creation of a master dictionary, the bag-of-words (BOW) technique can be used to represent each remaining word as a feature of the document. The preprocessing steps simplify the documents but also potentially degrade meaning. The values of the features in BOW can be modified by using techniques such as term count, term frequency, and term frequency-inverse document frequency. Experience and experimentation will guide decisions on which specific techniques will optimize ML performance. These and other NLP techniques are being applied in radiology. Radiologists' understanding of the strengths and limitations of these techniques will help in communication with data scientists and in implementation for specific tasks.

BMC medical informatics and decision making, 2017

The medical subdomain of a clinical note, such as cardiology or neurology, is useful content-derived metadata for developing machine learning downstream applications. To classify the medical subdomain of a note accurately, we have constructed a machine learning-based natural language processing (NLP) pipeline and developed medical subdomain classifiers based on the content of the note. We constructed the pipeline using the clinical NLP system, clinical Text Analysis and Knowledge Extraction System (cTAKES), the Unified Medical Language System (UMLS) Metathesaurus, Semantic Network, and learning algorithms to extract features from two datasets - clinical notes from Integrating Data for Analysis, Anonymization, and Sharing (iDASH) data repository (n = 431) and Massachusetts General Hospital (MGH) (n = 91,237), and built medical subdomain classifiers with different combinations of data representation methods and supervised learning algorithms. We evaluated the performance of classifier...

BMC Medical Informatics and Decision Making

Background Supervised machine learning algorithms have been a dominant method in the data mining field. Disease prediction using health data has recently shown a potential application area for these methods. This study aims to identify the key trends among different types of supervised machine learning algorithms, and their performance and usage for disease risk prediction. Methods In this study, extensive research efforts were made to identify those studies that applied more than one supervised machine learning algorithm on single disease prediction. Two databases (i.e., Scopus and PubMed) were searched for different types of search items. Thus, we selected 48 articles in total for the comparison among variants supervised machine learning algorithms for disease prediction. Results We found that the Support Vector Machine (SVM) algorithm is applied most frequently (in 29 studies) followed by the Naïve Bayes algorithm (in 23 studies). However, the Random Forest (RF) algorithm showed ...

… Intelligence for Modelling, …, 2005

Spam is commonly defined as unsolicited email messages and the goal of spam categorization is to distinguish between spam and legitimate email messages. Many researchers have been trying to separate spam from legitimate emails using machine learning algorithms based ...

Human Factors: The Journal of the Human Factors and Ergonomics Society, 2013

The objective of the paper is to understand leadership in an online community, specifically, Wikipedia. Background: Wikipedia successfully aggregates millions of volunteers' efforts to create the largest encyclopedia in human history. Without formal employment contracts and monetary incentives, one significant question for Wikipedia is how it organizes individual members with differing goals, experience, and commitment to achieve a collective outcome. Rather than focusing on the role of the small set of people occupying a core leadership position, we propose a shared leadership model to explain the leadership in Wikipedia. Members mutually influence one another by exercising leadership behaviors, including rewarding, regulating, directing, and socializing one another. Method: We conducted a two-phase study to investigate how distinct types of leadership behaviors (transactional, aversive, directive, and person-focused), the legitimacy of the people who deliver the leadership, and the experience of the people who receive the leadership influence the effectiveness of shared leadership in Wikipedia. Results: Our results highlight the importance of shared leadership in Wikipedia and identify trade-offs in the effectiveness of different types of leadership behaviors. Aversive and directive leadership increased contribution to the focal task, whereas transactional and person-focused leadership increased general motivation. We also found important differences in how newcomers and experienced members responded to leadership behaviors from peers. Application: These findings extend shared leadership theories, contribute new insight into the important underlying mechanisms in Wikipedia, and have implications for practitioners who wish to design more effective and successful online communities.

PloS one, 2011

E-communities, social groups interacting online, have recently become an object of interdisciplinary research. As with face-to-face meetings, Internet exchanges may not only include factual information but also emotional information--how participants feel about the subject discussed or other group members. Emotions in turn are known to be important in affecting interaction partners in offline communication in many ways. Could emotions in Internet exchanges affect others and systematically influence quantitative and qualitative aspects of the trajectory of e-communities? The development of automatic sentiment analysis has made large scale emotion detection and analysis possible using text messages collected from the web. However, it is not clear if emotions in e-communities primarily derive from individual group members' personalities or if they result from intra-group interactions, and whether they influence group activities.

Proceedings of the 7th International Conference on Advanced Data Mining and Applications Volume Part Ii, 2011

Retrieving relevant information from web, containing enormous amount of data, is a highly complicated research area. A landmark research that contributes to this area is web clustering which efficiently organizes a large amount of web documents into a small number of meaningful and coherent groups[1, 2]. Various techniques aim at accurately categorizing the web pages into clusters automatically. Suffix Tree Clustering (STC) is a phrase-based, state-of-art algorithm for web clustering that automatically groups semantically related documents based on shared phrases. Research has shown that it has outperformed other clustering algorithms such as K-means and Buckshot due to its efficient utilization of phrases to identify the clusters. Using STC as the baseline, we introduce a new method for ranking base clusters and new similarity measures for comparing clusters. Our STHAC technique combines the Heirarchical Agglomerative clustering method with phrase based Suffix Tree clustering to improve the cluster merging process. Experimental results have shown that STHAC outperforms the original STC as well as ESTC(our precious extended version of STC) with 16% increase in F-measure. This increase in F-measure of STHAC is achieved due to its better filtering of low score clusters, better similarity measures and efficient cluster merging algorithms.

Complexity

Are nearby places (e.g., cities) described by related words? In this article, we transfer this research question in the field of lexical encoding of geographic information onto the level of intertextuality. To this end, we explore Volunteered Geographic Information (VGI) to model texts addressing places at the level of cities or regions with the help of so-called topic networks. This is done to examine how language encodes and networks geographic information on the aboutness level of texts. Our hypothesis is that the networked thematizations of places are similar, regardless of their distances and the underlying communities of authors. To investigate this, we introduce Multiplex Topic Networks (MTN), which we automatically derive from Linguistic Multilayer Networks (LMN) as a novel model, especially of thematic networking in text corpora. Our study shows a Zipfian organization of the thematic universe in which geographical places (especially cities) are located in online communicati...

Health Information Science and Systems, 2017

Background: Patient healthcare trajectory is a recent emergent topic in the literature, encompassing broad concepts. However, the rationale for studying patients' trajectories, and how this trajectory concept is defined remains a public health challenge. Our research was focused on patients' trajectories based on disease management and care, while also considering medico-economic aspects of the associated management. We illustrated this concept with an example: a myocardial infarction (MI) occurring in a patient's hospital trajectory of care. The patient follow-up was traced via the prospective payment system. We applied a semi-automatic text mining process to conduct a comprehensive review of patient healthcare trajectory studies. This review investigated how the concept of trajectory is defined, studied and what it achieves. Methods: We performed a PubMed search to identify reports that had been published in peer-reviewed journals between January 1, 2000 and October 31, 2015. Fourteen search questions were formulated to guide our review. A semi-automatic text mining process based on a semantic approach was performed to conduct a comprehensive review of patient healthcare trajectory studies. Text mining techniques were used to explore the corpus in a semantic perspective in order to answer non-a priori questions. Complementary review methods on a selected subset were used to answer a priori questions. Results: Among the 33,514 publications initially selected for analysis, only 70 relevant articles were semi-automatically extracted and thoroughly analysed. Oncology is particularly prevalent due to its already well-established processes of care. For the trajectory thema, 80% of articles were distributed in 11 clusters. These clusters contain distinct semantic information, for example health outcomes (29%), care process (26%) and administrative and financial aspects (16%). Conclusion: This literature review highlights the recent interest in the trajectory concept. The approach is also gradually being used to monitor trajectories of care for chronic diseases such as diabetes, organ failure or coronary artery and MI trajectory of care, to improve care and reduce costs. Patient trajectory is undoubtedly an essential approach to be further explored in order to improve healthcare monitoring.

Health Informatics Journal

More and more health websites hire medical experts (physicians, medical students, experienced volunteers, etc.) and indicate explicitly their medical role in order to notify that they provide high-quality answers. However, medical experts may participate in forum discussions even when their role is not officially indicated. Detecting posts written by medical experts facilitates the quick access to posts that have more chances of being correct and informative. The main objective of this work is to learn classification models that can be used to detect posts written by medical experts in any health forum discussions. Two French health forums have been used to discover the best features and methods for this text categorization task. The obtained results confirm that models learned on appropriate websites may be used efficiently on other websites (more than 98% of F1-measure has been obtained using a Random Forest classifier). A study of misclassified posts highlights the participation ...

PLOS ONE

In this study we propose the use of text mining and machine learning methods to predict and detect Surgical Site Infections (SSIs) using textual descriptions of surgeries and post-operative patients' records, mined from the database of a high complexity University hospital. SSIs are among the most common adverse events experienced by hospitalized patients; preventing such events is fundamental to ensure patients' safety. Knowledge on SSI occurrence rates may also be useful in preventing future episodes. We analyzed 15,479 surgery descriptions and post-operative records testing different preprocessing strategies and the following machine learning algorithms: Linear SVC, Logistic Regression, Multinomial Naive Bayes, Nearest Centroid, Random Forest, Stochastic Gradient Descent, and Support Vector Classification (SVC). For prediction purposes, the best result was obtained using the Stochastic Gradient Descent method (79.7% ROC-AUC); for detection, Logistic Regression yielded the best performance (80.6% ROC-AUC).

Journal of Official Statistics

Every five years, the U.S. Census Bureau conducts the Economic Census, the official count of US businesses and the most extensive collection of data related to business activity. Businesses, policymakers, governments and communities use Economic Census data for economic development, business decisions, and strategic planning. The Economic Census provides key inputs for economic measures such as the Gross Domestic Product and the Producer Price Index. The Economic Census requires businesses to fill out a lengthy questionnaire, including an extended section about the goods and services provided by the business. To address the challenges of high respondent burden and low survey response rates, we devised a strategy to automatically classify goods and services based on product information provided by the business. We asked several businesses to provide a spreadsheet containing Universal Product Codes and associated text descriptions for the products they sell. We then used natural langu...

Scientific Programming

The ongoing growth in the vast amount of digital documents and other data in the Arabic language available online has increased the need for classification methods that can deal with the complex nature of such data. The classification of Arabic plays a large and important role in many modern applications and interferes with other sciences, which start from search engines and do not end with the Internet of Things. However, addressing the Arab classification errors with high performance is largely insufficient to deal with the huge quantities to reveal the classification of Arab documents; while some work was tackled out on the classification of the Arabic text, most of the research has focused on English text. The methods proposed for English are not suitable for Arabic as the morphology of the two languages differs substantially. Moreover, morphologically, the preprocessing of Arabic text is a particularly challenging task. In this study, three commonly used classification algorith...

Electronic Government, 2016

To conduct their business, organizations are nowadays challenged to handle huge amount of information from heterogeneous sources. Novel technologies can help them dealing with this delicate assignment. In this paper we describe an approach to document clustering and outlier detection that is regularly used to organize and summarize knowledge stored in huge amounts of documents in a government organization. The motivation for our preliminary study has been threefold: first, to obtain an overview of the topics addressed in the recently published e-government papers, with the emphasis on identifying the shift of focus through the years; second, to form a collection of papers related to a preselected terms of interest in order to explore the characteristic keywords that discriminate this collection with respect to the rest of the documents; and third, to compare the papers that address a similar topic from two document sources and to show characteristic similarities and differences between the two origins, with a particular aim to identify outlier papers in each document source that are potentially worth for further exploration. As a document source for our study we used E-Government Reference Library of articles and PubMed. The presented case study results suggest that the document exploration supported by a document clustering tool can be more focused, efficient and effective.

Applied Sciences, 2021

A core task in technology management in biomedical engineering and beyond is the classification of patents into domain-specific categories, increasingly automated by machine learning, with the fuzzy language of patents causing particular problems. Striving for higher classification performance, increasingly complex models have been developed, based not only on text but also on a wealth of distinct (meta) data and methods. However, this makes it difficult to access and integrate data and to fuse distinct predictions. Although the already established Cooperate Patent Classification (CPC) offers a plethora of information, it is rarely used in automated patent categorization. Thus, we combine taxonomic and textual information to an ensemble classification system comparing stacking and fixed combination rules as fusion methods. Various classifiers are trained on title/abstract and on both the CPC and IPC (International Patent Classification) assignments of 1230 patents covering six categ...

Proceedings of the 18th ACM/IEEE on Joint Conference on Digital Libraries

We develop an evaluation framework for the validation of conformance checkers for the long-term preservation. The framework assesses the correctness, usability, and usefulness of the tools for

Suicide & life-threatening behavior, 2016

Death by suicide demonstrates profound personal suffering and societal failure. While basic sciences provide the opportunity to understand biological markers related to suicide, computer science provides opportunities to understand suicide thought markers. In this novel prospective, multimodal, multicenter, mixed demographic study, we used machine learning to measure and fuse two classes of suicidal thought markers: verbal and nonverbal. Machine learning algorithms were used with the subjects' words and vocal characteristics to classify 379 subjects recruited from two academic medical centers and a rural community hospital into one of three groups: suicidal, mentally ill but not suicidal, or controls. By combining linguistic and acoustic characteristics, subjects could be classified into one of the three groups with up to 85% accuracy. The results provide insight into how advanced technology can be used for suicide assessment and prevention.

2017

We introduce the notion of “seminar users”, who are social media users engaged in propaganda in support of a political entity. We develop a framework that can identify such users with 84.4% precision and 76.1% recall. While our dataset is from the Arab region, omitting language-specific features has only a minor impact on classification performance, and thus, our approach could work for detecting seminar users in other parts of the world and in other languages. We further explored a controversial political topic to observe the prevalence and potential potency of such users. In our case study, we found that 25% of the users engaged in the topic are in fact seminar users and their tweets make nearly a third of the on-topic tweets. Moreover, they are often successful in affecting mainstream discourse with coordinated hashtag campaigns.

Lecture Notes in Geoinformation and Cartography, 2016

Semantically rich 3D building models give the potential for a wealth of rich geo-spatially-enabled applications such as cultural heritage augmented reality, urban planning, radio network planning and personal navigation. However, the majority of existing building models lack much if any semantic detail. This work demonstrates a novel method for automatically locating subclasses of windows and doors, using computer vision techniques including the histogram of oriented gradient (HOG) template matching, and automatically creating enriched CityGML content for the matched windows and doors. Good results were achieved for class identification with potential for further refinement of subclasses of windows and doors and other architectural features. It is part of a wider project to bring even richer semantic content to 3D geo-spatial building models.

Scientific Programming

In this paper, a novel approach is presented for authorship identification in English and Urdu text using the LDA model with n-grams texts of authors and cosine similarity. The proposed approach uses similarity metrics to identify various learned representations of stylometric features and uses them to identify the writing style of a particular author. The proposed LDA-based approach emphasizes instance-based and profile-based classifications of an author’s text. Here, LDA suitably handles high-dimensional and sparse data by allowing more expressive representation of text. The presented approach is an unsupervised computational methodology that can handle the heterogeneity of the dataset, diversity in writing, and the inherent ambiguity of the Urdu language. A large corpus has been used for performance testing of the presented approach. The results of experiments show superiority of the proposed approach over the state-of-the-art representations and other algorithms used for authors...

Neural Computing and Applications, 2021

Risk management plays a vital role in tackling cyber threats within the cyber-physical system (CPS). It enables identifying critical assets, vulnerabilities and threats and determining suitable proactive control measures for the risk mitigation. However, due to the increased complexity of the CPS, cyber-attacks nowadays are more sophisticated and less predictable, which makes risk management task more challenging. This paper aims for an effective cybersecurity risk management (CSRM) practice using assets criticality, predication of risk types and evaluating the effectiveness of existing controls. We follow a number of techniques for the proposed unified approach including fuzzy set theory for the asset criticality, machine learning classifiers for the risk predication and comprehensive assessment model (CAM) for evaluating the effectiveness of the existing controls. The proposed approach considers relevant CSRM concepts such as asset, threat actor, attack pattern, tactic, technique ...

2018

Recent machine learning approaches for classifying text as human-written or bot-generated rely on training sets that are large, labeled diligently, and representative of the underlying domain. While valuable, these machine learning approaches ignore programs as an additional source of such training sets. To address this problem of incomplete training sets, this paper proposes to systematically supplement existing training sets with samples inferred via program analysis. In our preliminary evaluation, training sets enriched with samples inferred via dynamic symbolic execution were able to improve machine learning classifier accuracy for simple string-generating programs.

The vast amount of electronic documents available on the Internet demands for automatic tools that help people finding, organizing and easily accessing to all this information. Although current text classification methods have alleviated some of the above problems, such strategies depend on having a large and reliable set of labeled data. In order to overcome such limitation, this work proposes an alternative approach for semi-supervised text classification, which is based on a new strategy for diminishing the sensitivity to the noise contained on labeled data by means of automatic text summarization. Experimental results showed that our proposed approach outperforms traditional semi-supervised text classification techniques; additionally, our results also indicate that our approach is suitable for learning from only one labeled example per category.

Proceedings of the 29th Annual ACM Symposium on Applied Computing - SAC '14, 2014

Lately, organizations have been subject to regulation promoting information transparency; one example of this is the Brazilian Information Access Law. This paper presents a novel way of performing requirements elicitation using both the law and a Non-Functional Requirements Patterns catalog as the information sources. Since organizations must follow the law, its information systems must also implement the law as requirements. Our process is guided by pattern matching, text mining and grounded analysis. We examine the special case of the Brazilian Access Law using our approach, which compares a previously encoded transparency knowledge base with the law.

Proceedings of the ACM/IEEE Joint Conference on Digital Libraries in 2020, 2020

Web archive data usually contains high-quality documents that are very useful for creating specialized collections of documents, e.g., scientific digital libraries and repositories of technical reports. In doing so, there is a substantial need for automatic approaches that can distinguish the documents of interest for a collection out of the huge number of documents collected by web archiving institutions. In this paper, we explore different learning models and feature representations to determine the best performing ones for identifying the documents of interest from the web archived data. Specifically, we study both machine learning and deep learning models and "bag of words" (BoW) features extracted from the entire document or from specific portions of the document, as well as structural features that capture the structure of documents. We focus our evaluation on three datasets that we created from three different Web archives. Our experimental results show that the BoW...

2020

In this paper, we show how selecting and combining encodings of natural and mathematical language affect classification and clustering of documents with mathematical content. We demonstrate this by using sets of documents, sections, and abstracts from the arXiv preprint server that are labeled by their subject class (mathematics, computer science, physics, etc.) to compare different encodings of text and formulae and evaluate the performance and runtimes of selected classification and clustering algorithms. Our encodings achieve classification accuracies up to 82.8% and cluster purities up to 69.4% (number of clusters equals number of classes), and 99.9% (unspecified number of clusters) respectively. We observe a relatively low correlation between text and math similarity, which indicates the independence of text and formulae and motivates treating them as separate features of a document. The classification and clustering can be employed, e.g., for document search and recommendation...

Advances in Science, Technology and Engineering Systems Journal, 2020

Article history: Received: 8 July, 2020 Accepted: 11 August, 2020 Online: 28 August, 2020

The International Conference on Advanced Machine Learning Technologies and Applications (AMLTA2018)

with the recent explosive growth of e-commerce, and online communication, a new kind of text, opinion text become available in many application areas, such as Instant Messages, online Chat and review websites. However, Opinion mining is developed simultaneously therewith then used for product feedback analysis, and for decision making to users and companies. According to these big needs we realized a tool for the benefit of the Opinion Mining and Sentiments Analysis. This tool is based on the combination of both SVM machine learning algorithm and approaches in terms of extraction opinion text features. Our tool collects data in real time from (Amazon, Cnet and TrustedReviews) websites according to user request product. After it filters reviews from the other content then extracting opinions (subjective sentences) and classifies them. Moreover, to improve the performance of our system we proposed some algorithms that constructed on sentiment bag, based on emoticons and injections.

Advances in Astronomy

Pulsar stars, usually neutron stars, are spherical and compact objects containing a large quantity of mass. Each pulsar star possesses a magnetic field and emits a slightly different pattern of electromagnetic radiation which is used to identify the potential candidates for a real pulsar star. Pulsar stars are considered an important cosmic phenomenon, and scientists use them to study nuclear physics, gravitational waves, and collisions between black holes. Defining the process of automatic detection of pulsar stars can accelerate the study of pulsar stars by scientists. This study contrives an accurate and efficient approach for true pulsar detection using supervised machine learning. For experiments, the high time-resolution (HTRU2) dataset is used in this study. To resolve the data imbalance problem and overcome model overfitting, a hybrid resampling approach is presented in this study. Experiments are performed with imbalanced and balanced datasets using well-known machine learn...

Multimedia Tools and Applications

Social media platforms are becoming a rich source of valuable information through sharing and publishing user generated reviews and comments. The identification and extraction of subjective information from a piece of text is a crucial challenge in sentiment analysis. Numerous techniques have been proposed that aimed to analyze the sentiments of the text. However, accuracy was compromised due to inadequate identification of intensity, sentiments shifter and negation of words as well as divergence between tweets and their associated labels. In this study, Prescriptive Sentiment Analysis (PSA) based on features synchronization has been introduced for increasing accuracy in text sentiment analysis. At first, pre-processing has been performed which includes removal of stop words and tokenizing the text into sentiment words, intensity words, sentiment shifters and negation words. Secondly; polarity of intensity words, clauses and sentiment shifters in the text are calculated. Identification and removal of ambiguity between extracted features and their associated labels have been accomplished through feature synchronization. The K-Nearest Neighbor (KNN) has been implemented to predict text trend based on synchronized features. The proposed approach has been evaluated on publicly available datasets of twitter and movie reviews. Experimental results show significant improvement in sentiment analysis efficiency as compared to other baseline methods.

PeerJ Computer Science

The coronavirus disease 2019 (COVID-19) causes tremendous damages to the world, including threats to human’s health and daily activities. Most industries have been affected by this pandemic, particularly the tourism industry. The online travel agencies (OTAs) have suffered from the global tourism market crisis by air travel lockdown in many countries. How online travel agencies can survive at stake and prepare for the post-COVID-19 future has emerged as an urgent issue. This study aims to examine the critical factors of customers’ satisfaction to OTAs during the COVID-19 pandemic. A text mining method for feature selection, namely LASSO, was used to deal with online customer reviews and to extract factors that shape customers’ satisfaction to OTAs. Results showed that refunds, promptness, easiness and assurance were ranked as the most competitive factors of customers’ satisfaction, followed by bad reviews & cheap and excellent service & comparison. New factors to customers’ satisfac...

Engineering Reports

In this work a novel framework for modeling role and task allocation in Cooperative Heterogeneous Multi-Robot Systems (CHMRSs) is presented. This framework encodes a CHMRS as a set of multi-dimensional relational structures (MDRSs). This set of structure defines collaborative tasks through both temporal and spatial relations between processes of heterogeneous robots. These relations are enriched with tensors which allow for geometrical reasoning about collaborative tasks. A learning schema is also proposed in order to derive the components of each MDRS. According to this schema, the components are learnt from data reporting the situated history of the processes executed by the team of robots. Data are organized as a Multi-Robot Collaboration Treebank (MRCT) in order to support learning. Moreover, a generative approach, based on a probabilistic model, is combined together with Non-Negative Tensor Decomposition (NTD) for both building the tensors and estimating latent knowledge. Preliminary evaluation of the performance of this framework is performed in simulation with three heterogeneous robots, namely, two Unmanned Ground Vehicles (UGVs) and one Unmanned Aerial Vehicle (UAV).

Lecture Notes in Computer Science, 2007

A major problem encountered by text clustering practitioners is the difficulty of determining a priori which is the optimal text representation and clustering technique for a given clustering problem. As a step towards building robust document partitioning systems, we present a strategy based on a hierarchical consensus clustering architecture that operates on a wide diversity of document representations and partitions. The conducted experiments show that the proposed method is capable of yielding a consensus clustering that is comparable to the best individual clustering available even in the presence of a large number of poor individual labelings, thus largely avoiding the problems derived from selecting non-optimal clustering configurations.

Healthcare

COVID-19 pandemic has caused a global health crisis, resulting in endless efforts to reduce infections, fatalities, and therapies to mitigate its after-effects. Currently, large and fast-paced vaccination campaigns are in the process to reduce COVID-19 infection and fatality risks. Despite recommendations from governments and medical experts, people show conceptions and perceptions regarding vaccination risks and share their views on social media platforms. Such opinions can be analyzed to determine social trends and devise policies to increase vaccination acceptance. In this regard, this study proposes a methodology for analyzing the global perceptions and perspectives towards COVID-19 vaccination using a worldwide Twitter dataset. The study relies on two techniques to analyze the sentiments: natural language processing and machine learning. To evaluate the performance of the different lexicon-based methods, different machine and deep learning models are studied. In addition, for s...

Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval

BMC medical informatics and decision making, 2015

Narrative text is a useful way of identifying injury circumstances from the routine emergency department data collections. Automatically classifying narratives based on machine learning techniques is a promising technique, which can consequently reduce the tedious manual classification process. Existing works focus on using Naive Bayes which does not always offer the best performance. This paper proposes the Matrix Factorization approaches along with a learning enhancement process for this task. The results are compared with the performance of various other classification approaches. The impact on the classification results from the parameters setting during the classification of a medical text dataset is discussed. With the selection of right dimension k, Non Negative Matrix Factorization-model method achieves 10 CV accuracy of 0.93.

2018

Focused crawlers aim to automatically discover online content resources relevant to a domain of interest by automatically navigating through the Web link structure and selecting which hyperlinks to follow based on an estimation of their relevance to the topic of interest; to this end, classifier-guided approaches are typically employed for identifying hyperlinks having the higher likelihood of leading to relevant content. However, the training data used for building these classifiers might not be entirely representative of the domain of interest, or the domain of interest might change over time. To meet these challenges, this work proposes a novel adaptive focused crawling framework that allows the classifiers that underlie the hyperlink selection policy to be adapted based on the evidence they encounter during their crawls. Our framework uses two different approaches to retrain its models: (i) Interactive Adaptation, where a user manually evaluates the discovered resources, and (ii...

Evolution in Computational Intelligence

Training a machine learning model always requires an ample quantity of data as the effectiveness of the model depends on how good it is trained. It is not always that easy, using Zero-shot learning this paper is concentrated on intent classification in different speeches given the model is trained only in one language. Several machine learning models were developed and their outputs were compared to understand the concept of how Zero-shot learning (ZSL) works. It is grounded on the idea that the sentence embedding of two sentences of different languages having the same sense must be similar. In order to predict the intent of sentences in different languages, every sentence must be signified in the same manner and in the same vector space. Here comes the Language Agnostic Sentence Representations or LASER which uses a single encoder to give sentence embedding vector for sentences in any language. The motive of this paper is to assess how useful these sentence representations are in the context of a particular scenario.

International Journal of Advanced Pervasive and Ubiquitous Computing

Recommender systems have been used successfully in order to deal with information overload problems in a wide variety of domains ranging from e-commerce, e-tourism, to e-learning. They typically predict the ratings of unseen items by a user and recommend the top N items based on user's profile. Moreover, the profile can be enriched further by using additional information such as contextual data, domain knowledge, and tagging information among others for improving the quality of recommendations. Traditional approaches have not been effective in exploiting these additional data sources. Hence, new techniques need to be developed for extracting and integrating them into the recommendation process. In this article, the authors present a survey on state of the art recommendation approaches their algorithms, issues and also provides further research directions for developing smart and intelligent recommender systems.

Journal of Quantitative Linguistics, 2010

This paper provides a century-wide quantitative analysis of the Turkish literature using forty novels of forty authors. We divide the century into four eras or quarter centuries; allocate ten novels to each era, and partition each novel into equal sized blocks. Using cross validation based discriminant analysis, with the most frequent words as discriminators; we achieve a classification rate with a relatively high accuracy when the novel blocks are classified according to their eras. We show that, by using statistical stylistic methods, the author gender of Turkish text can be accurately identified. We also study the gender differences regarding the use of most frequent words. Using weighted least squares regression and a sliding window approach we show that as time passes, words, both in terms of tokens (in text) and types (in vocabulary), have become longer. The findings of this work have implications for the historical linguistic analysis of the Turkish language.

Arabian Journal for Science and Engineering

Advances in Usability, User Experience and Assistive Technology

Few years back, women harassment was not an issue. Nowadays, women harassment is recognize as a legitimate human rights issue and threat to women's well-being. The #MeToo campaign launched by actor Alyssa Milano to drew the world's attention to women harassment, with a hashtag that went viral. Harassment is unwanted actions and comments by strangers. Due to the fear of harassment, women are paying to stay safe as women don't walk independently as men. Pakistani women were among of them. This research will illustrate the harassment problems faced by working women and will examine the available solutions with the context of Pakistani women. There is no evidence that these apps have the power to decrease incidents, despite what the apps developers' claim. This research will also examine these apps to find out why women still do not feel safer as the apps claims. Keywords: #MeToo Á Safety apps Á Women centered design Á Women safety Usability Á Design Á User experience

Journal of Veterinary Diagnostic Investigation, 2017

The histologic evaluation of gastrointestinal (GI) biopsies is the standard for diagnosis of a variety of GI diseases (e.g., inflammatory bowel disease [IBD] and alimentary lymphoma [ALA]). The World Small Animal Veterinary Association (WSAVA) Gastrointestinal International Standardization Group proposed a reporting standard for GI biopsies consisting of a defined set of microscopic features. We compared the machine classification accuracy of free-text microscopic findings with those represented in the WSAVA format with a diagnosis of IBD and ALA. Unstructured free-text duodenal biopsy pathology reports from cats ( n = 60) with a diagnosis of IBD ( n = 20), ALA ( n = 20), or normal ( n = 20) were identified. Biopsy samples from these cases were then scored following the WSAVA guidelines to create a set of structured reports. Three supervised machine-learning algorithms were trained using the structured and then the unstructured reports. Diagnosis classification accuracy for the 3 al...

Lecture Notes in Computer Science, 2005

This paper describes a supervised three-tier clustering method for classifying students' essays of qualitative physics in the Why2-Atlas tutoring system. Our main purpose of categorizing text in our tutoring system is to map the students' essay statements into principles and misconceptions of physics. A simple `bag-of-words' representation using a naïve-bayes algorithm to categorize text was unsatisfactory for our purposes of analyses as it exhibited many misclassifications because of the relatedness of the concepts themselves and its inability to handle misconceptions. Hence, we investigate the performance of the k-nearest neighborhood algorithm coupled with clusters of physics concepts on classifying students' essays. We use a three-tier tagging schemata (cluster, sub-cluster and class) for each document and found that this kind of supervised hierarchical clustering leads to a better understanding of the student's essay.

Lecture Notes in Computer Science, 2011

This paper explores how background knowledge from freely available web resources can be utilised for Textual Case Based Reasoning. The work reported here extends the exisiting Explicit Semantic Analysis approach to representation, where textual content is represented using concepts with correspondence to Wikipedia articles. We present approaches to identify Wikipedia pages that are likely to contribute to the effectiveness of text classification tasks. We also study the effect of modelling semantic similarity between concepts (amounting to Wikipedia articles) empirically. We conclude with the observation that integrating background knowledge from resources like Wikipedia into TCBR tasks holds a lot of promise as it can improve system effectiveness even without elaborate manual knowledge engineering. Significant performance gains are obtained using a very small number of features that have very strong correspondence to how humans describe the domain.

Encyclopedia of Social Network Analysis and Mining, 2017

Sentiment Analysis: This is text analysis aiming to determine the attitude of a speaker or a writer with respect to some topic or the overall contextual polarity of a piece of text. Sentiment analysis on Twitter is the use of natural language processing techniques to identify and categorize opinions expressed in a tweet, in order to determine the author's attitude toward a particular topic or in general. Typically, discrete labels such as positive, negative, neutral, and objective are used for this purpose, but it is also possible to use labels on an ordinal scale, or even continuous numerical values.

SN Applied Sciences

Machine Learning has found application in solving complex problems in different fields of human endeavors such as intelligent gaming, automated transportation, cyborg technology, environmental protection, enhanced health care, innovation in banking and home security, and smart homes. This research is motivated by the need to explore the global structure of machine learning to ascertain the level of bibliographic coupling, collaboration among research institutions, co-authorship network of countries, and sources coupling in publications on machine learning techniques. The Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) was applied to clustering prediction of authors dominance ranking in this paper. Publications related to machine learning were retrieved and extracted from the Dimensions database with no language restrictions. Bibliometrix was employed in computation and visualization to extract bibliographic information and perform a descriptive ana...

Natural Language Engineering, 2017

Natural language processing (NLP) techniques can be used to make inferences about peoples’ mental states from what they write on Facebook, Twitter and other social media. These inferences can then be used to create online pathways to direct people to health information and assistance and also to generate personalized interventions. Regrettably, the computational methods used to collect, process and utilize online writing data, as well as the evaluations of these techniques, are still dispersed in the literature. This paper provides a taxonomy of data sources and techniques that have been used for mental health support and intervention. Specifically, we review how social media and other data sources have been used to detect emotions and identify people who may be in need of psychological assistance; the computational techniques used in labeling and diagnosis; and finally, we discuss ways to generate and personalize mental health interventions. The overarching aim of this scoping revi...

Social Computing and Social Media. Applications and Analytics, 2017

Twitter is a microblogging platform that allows users to share opinions with a restricted amount of characters. Given the social characteristic of Twitter, it is a potential source for sentiment analysis. For this reason, opinions of certain subjects such as people or brands can change in short periods of time. A traditional approach of a sentiment classifier implementation performs poorly since it depends on how it is trained. We propose a novel method for tackling this problem, with the implementation of a multi-agent system for classifying and corpus analysis mechanism for retraining the classifier. This mechanism consists of a critic agent which compares the trained corpus of the classifier agent with new collections of documents from future time steps, using primarily two methods: hypothesis test analysis and histogram differences. A Naïve-Bayes based classifier was implemented with this mechanism with multiple configurations. The results of experimental data show that the mechanism boosts its performance, when compared to a pure Naïve Bayes classifier.

Recommender System with Machine Learning and Artificial Intelligence, 2020

This paper foregrounds the Native Hawai'ians' environmental perspective of Aloha Aina in Oswald Andrew's Ka'a'awa. By asserting that the land and all the entities are part of their family (ohana), this epistemology stresses the equality of human and non-human which run contrary to the Western anthropocentric view. This present study is conducted within indigenous ecocriticism perspective, in which alternative epistemology of human and non-human interaction should be considered in the light of global environmental crisis. In Ka'a'awa, Bushnell explores how reconciliation of indigenous perspective and White settler proves difficult to achieve under Orientalist discourse that derogatively perceive the native epistemologies as superstitious and irrational. Moreover, the enforcement of Western anthropocentric view towards the local natives through colonialism causes the islanders to forget their ancestral epistemology. This paper concludes that Bushnell's elaboration of a novel culture which embraces both side of the spectrum, indigenous and white settler culture is an avenue to achieve sustainable ecological condition.

Springer Proceedings in Business and Economics, 2016

The purpose of this paper is to investigate the application of text classification in Hypatia, the digital library of Technological Educational Institute of Athens, in order to provide an automated classification tool as an alternative to manual assignments. The crucial point in text classification is the selection of the most important term-words for document representation. Classic weighting method TF.IDF was investigated. Our document collection consists of 718 abstracts in Medicine, Tourism and Food Technology. Classification was conducted utilizing 14 classifiers available on WEKA. Classification process yielded an excellent ~97% precision score.

Corporate Reputation Review, 2017

The reputation literature tends to approach the relationship between firms' reputation and performance as a direct causal relationship. To improve our understanding of corporate reputation it is necessary to re-conceptualise reputation as having an indirect influence. Reputation acts as a source of resilience providing firms with an enhanced ability to adapt when faced with external difficulties as well as allowing the firm to rebound following a performance decline. Rather than relying on traditional measures, this research analysed the content of 10,582 Australian firms' annual reports over 17 years to identify what executives communicate to their stakeholders. These results were then subjected to analyses which identified that firms with superior financial and service reputation tended to outperform firms without such standing.

Complexity, 2019

Nowadays, smart devices as a part of daily life collect data about their users with the help of sensors placed on them. Sensor data are usually physical data but mobile applications collect more than physical data like device usage habits and personal interests. Collected data are usually classified as personal, but they contain valuable information about their users when it is analyzed and interpreted. One of the main purposes of personal data analysis is to make predictions about users. Collected data can be divided into two major categories: physical and behavioral data. Behavioral data are also named as neurophysical data. Physical and neurophysical parameters are collected as a part of this study. Physical data contains measurements of the users like heartbeats, sleep quality, energy, movement/mobility parameters. Neurophysical data contain keystroke patterns like typing speed and typing errors. Users’ emotional/mood statuses are also investigated by asking daily questions. Six...

Celal Bayar Üniversitesi Fen Bilimleri Dergisi, 2021

Identifying the authors of a given set of text is a well addressed and complicated task. It requires thorough knowledge of different authors' writing styles and discriminating them. As the main contribution of this paper, we propose to perform this task using machine learning and deep learning methods, state-of-the-art algorithms, and methods used in numerous complex Natural Language Processing (NLP) problems. We used a text corpus of daily newspaper columns written by thirty authors to perform our experiments. The experimental results proved that document embeddings trained via neural network architecture achieve cutting edge accuracy in learning writing styles and identifying authors of given writings even though the dataset has a considerably unbalanced distribution. We represent our experimental results and outsource our codes for interested readers and natural language processing (NLP) enthusiasts as a GitHub repository. They can reproduce and confirm the results and modify them according to their own needs.

Health informatics journal, 2017

Processing huge repository of medical literature for extracting relevant and high-quality evidences demands efficient evidence support methods. We aim at developing methods to automate the process of finding quality evidences from a plethora of literature documents and grade them according to the context (local condition). We propose a two-level methodology for quality recognition and grading of evidences. First, quality is recognized using quality recognition model; second, context-aware grading of evidences is accomplished. Using 10-fold cross-validation, the proposed quality recognition model achieved an accuracy of 92.14 percent and improved the baseline system accuracy by about 24 percent. The proposed context-aware grading method graded 808 out of 1354 test evidences as highly beneficial for treatment purpose. This infers that around 60 percent evidences shall be given more importance as compared to the other 40 percent evidences. The inclusion of context in recommendation of ...

Scientometrics, 2008

Recently, philosophers of science have argued that the epistemological requirements of different scientific fields lead necessarily to differences in scientific method. In this paper, we examine possible variation in how language is used in peer-reviewed journal articles from various fields to see if features of such variation may help to elucidate and support claims of methodological variation among the sciences. We hypothesize that significant methodological differences will be reflected in related differences in scientists' language style. This paper reports a corpus-based study of peer-reviewed articles from twelve separate journals in six fields of experimental and historical sciences. Machine learning methods were applied to compare the discourse styles of articles in different fields, based on easily-extracted linguistic features of the text. Features included function word frequencies, as used often in computational stylistics, as well as lexical features based on systemic functional linguistics, which affords rich resources for comparative textual analysis. We found that indeed the style of writing in the historical sciences is readily distinguishable from that of the experimental sciences. Furthermore, the most significant linguistic features of these distinctive styles are directly related to the methodological differences posited by philosophers of science between historical and experimental sciences, lending empirical weight to their contentions.

International Journal of Information Retrieval Research, 2022

Software practitioners often strive to achieve a “bug-free” software, though, it is a myth. Software Bug Categorization (SBC) models, which assigns levels (viz. “low”, “moderate” or “high”) to a software bug aid effective bug management. They assist in allocation of proper maintenance resources for bug elimination to improve software quality. This study proposes the development of SBC models that allocate levels on the basis of three software bug aspects i.e., maintenance effort required to correct a bug, its change impact and the combined effect of both of these. In order to develop SBC models, we use text mining approach, which extracts relevant features from bug descriptions and relates these features with different software bug levels. The results of the study indicate that the categorization of software bugs in accordance with maintenance effort and change impact is possible. Furthermore, the combined approach SBC models were also found to be effective.

Proceedings of the 2015 International Conference on The Theory of Information Retrieval, 2015

In multi-class text classification, the performance (effectiveness) of a classifier is usually measured by micro-averaged and macro-averaged F1 scores. However, the scores themselves do not tell us how reliable they are in terms of forecasting the classifier's future performance on unseen data. In this paper, we propose a novel approach to explicitly modelling the uncertainty of average F1 scores through Bayesian reasoning.

Advances in Intelligent Systems and Computing, 2015

In this paper, a supervised automatic text documents classification using the fuzzy decision trees technique is proposed. Whatever the algorithm used in the fuzzy decision trees, there must be a criterion for the choice of discriminating attribute at the nodes to partition. For fuzzy decision trees two heuristics are usually used to select the discriminating attribute at the node to partition. In the field of text documents classification there is a heuristic that has not yet been tested. This paper tested this heuristic.

PeerJ, 2017

Introduction Assessment of episodic memory has been traditionally used to evaluate potential cognitive impairments in senior adults. Typically, episodic memory evaluation is based on personal interviews and pen-and-paper tests. This article presents the design, development and a preliminary validation of a novel digital game to assess episodic memory intended to overcome the limitations of traditional methods, such as the cost of its administration, its intrusive character, the lack of early detection capabilities, the lack of ecological validity, the learning effect and the existence of confounding factors. Materials and Methods Our proposal is based on the gamification of the California Verbal Learning Test (CVLT) and it has been designed to comply with the psychometric characteristics of reliability and validity. Two qualitative focus groups and a first pilot experiment were carried out to validate the proposal. Results A more ecological, non-intrusive and better administrable to...

Language Resources and Evaluation, 2012

Web 2.0 provides user-friendly tools that allow persons to create and publish content online. User generated content often takes the form of short texts (e.g., blog posts, news feeds, snippets, etc). This has motivated an increasing interest on the analysis of short texts and, specifically, on their categorisation. Text categorisation is the task of classifying documents into a certain number of predefined categories. Traditional text classification techniques are mainly based on word frequency statistical analysis and have been proved inadequate for the classification of short texts where word occurrence is too small. On the other hand, the classic approach to text categorization is based on a learning process that requires a large number of labeled training texts to achieve an accurate performance. However labeled documents might not be available, when unlabeled documents can be easily collected. This paper presents an approach to text categorisation which does not need a preclassified set of training documents. The proposed method only requires the category names as user input. Each one of these categories is defined by means of an ontology of terms modelled by a set of what we call proximity equations. Hence, our method is not category occurrence frequency based, but highly depends on the definition of that category and how the text fits that definition. Therefore, the proposed approach is an appropriate method for short text classification where the frequency of occurrence of a category is very small or even zero. Another feature of our method is that the classification process is based on the ability of an extension of the standard Prolog language, named Bousi∼Prolog, for flexible matching and knowledge representation. This declarative approach provides a text classifier which is quick and easy to build, and a classification process which is easy for the user to understand. The results of experiments showed that the proposed method achieved a reasonably useful performance.

Sensors (Basel, Switzerland), 2018

Wearable sensors for human physiological monitoring have attracted tremendous interest from researchers in recent years. However, most of the research involved simple trials without any significant analytical algorithms. This study provides a way of recognizing human motion by combining textile stretch sensors based on single-walled carbon nanotubes (SWCNTs) and spandex fabric (PET/SP) and machine learning algorithms in a realistic application. In the study, the performance of the system will be evaluated by identification rate and accuracy of the motion standardized. This research aims to provide a realistic motion sensing wearable product without unnecessary heavy and uncomfortable electronic devices.

Lecture Notes in Computer Science, 2009

In this paper, we introduce a Gene Expression Programming algorithm for multi label classification. This algorithm encodes each individual into a discriminant function that shows whether a pattern belongs to a given class or not. The algorithm also applies a niching technique to guarantee that the population includes functions for each existing class. In order to evaluate the quality of our algorithm, its performance is compared to that of four recently published algorithms. The results show that our proposal is the best in terms of accuracy, precision and recall.

Applied Sciences, 2020

'El Diario de Juárez' is a local newspaper in a city of 1.5 million Spanish-speaking inhabitants that publishes texts of which citizens read them on both a website and an RSS (Really Simple Syndication) service. This research applies natural-language-processing and machine-learning algorithms to the news provided by the RSS service in order to classify them based on whether they are about a traffic incident or not, with the final intention of notifying citizens where such accidents occur. The classification process explores the bag-of-words technique with five learners (Classification and Regression Tree (CART), Naïve Bayes, kNN, Random Forest, and Support Vector Machine (SVM)) on a class-imbalanced benchmark; this challenging issue is dealt with via five sampling algorithms: synthetic minority oversampling technique (SMOTE), borderline SMOTE, adaptive synthetic sampling, random oversampling, and random undersampling. Consequently, our final classifier reaches a sensitivity of 0.86 and an area under the precision-recall curve of 0.86, which is an acceptable performance when considering the complexity of analyzing unstructured texts in Spanish.

Intelligent Data Analysis, 2019

Due to the rapid growth of web platforms such as blogs, discussion forums, peer-to-peer networks, and various other types of social media, Sentiment Polarity Detection (SPD) (classifying texts by "positive" or "negative" orientation) has become more important and challenging task in recent years. There is a growing need for management and study of SPD not only in English, but also in other languages. The key reason for using Machine Learning (ML) for SPD lies in engineering a representative set of features. This paper explores different (byte, character and word) n-gram based text representation models in order to determine the most valuable model for the representation of text documents in various languages, which can be used successfully by ML classification techniques for solving SPD task. Proposed n-gram models were used in conjunction with k Nearest Neighbourhood (kNN), Support Vector Machine (SVM) and Maximum Entropy (MaxEnt) algorithms to determine opinion polarity of the proposed movie reviews. The effectiveness and language independence of the proposed n-gram models were demonstrated in experiments performed on seven publicly available movie review benchmarks in Arabic, Czech, English, French, Spanish,Turkish, and Serbian being the authors' mother tongue. Formal evaluation has confirmed that the proposed byte and character n-gram models outperform word n-gram model, and in conjunction with the presented MaxEnt algorithm outperform other ML supervised techniques used with more complex document representation approaches. In some cases (Arabic, Czech, French, Serbian and Turkish), signficant improvements over the baselines have been achieved. Despite their simplicity and broad applicability, byte and character n-grams have been shown to be able to capture information on different levels-lexical and syntactic.

ACM Transactions on Information Systems, 2007

Web search engines typically provide search results without considering user interests or context. We propose a personalized search approach that can easily extend a conventional search engine on the client side. Our mapping framework automatically maps a set of known user interests onto a group of categories in the Open Directory Project (ODP) and takes advantage of manually edited data available in ODP for training text classifiers that correspond to, and therefore categorize and personalize search results according to user interests. In two sets of controlled experiments, we compare our personalized categorization system (PCAT) with a list interface system (LIST) that mimics a typical search engine and with a nonpersonalized categorization system (CAT). In both experiments, we analyze system performances on the basis of the type of task and query length. We find that PCAT is preferable to LIST for information gathering types of tasks and for searches with short queries, and PCAT ...

Neural Computing and Applications

Manufacturers customize computer components upon receipt of sale orders. They perform burn-in tests on each unit of product before shipment to ensure a high standard of quality. Burn-in is normally associated with high production costs and slows down manufacturing operations. This study aims to enhance the manufacturing process by predicting test failure patterns using machine learning methods. By identifying the components that are likely to cause failures, manufacturers can accelerate the rectification process and improve delivery time which in turn leads to better customer service. This study hypothesized that the component of concern produces a higher test failure rate. To provide insight into the data and test the hypothesis, descriptive and predictive analytics are used at various stages. Predictive analytics was performed using machine learning via Naïve Bayes since it outperformed SVM and Random Forest classifier. For the descriptive analysis stage, a visual representation revealed many components (81) to be associated with a more than average test failure rate. Fisher's exact test confirmed that 12 of them are statistically significant and worth studying their behaviour further. Moreover, an association rule mining exercise identified several combinations of modules that have a higher inclination with the test failure. For the predictive analytics stage, the Naïve Bayes classifier predicted test failure with 79% accuracy and 53% recall rate. Another Naïve Bayes classifier predicted error messages associated with a test failure with 68% recall rate over manually labelled error messages. However, a neural network-based automatic text classifier was developed and tested that yielded 66% accuracy. This analysis provides the foundation for a recommendation made that can reduce the burn test failure rate by 25% which is expected to increase further with the improved performance model upon training with a larger data set.

Journal of Intelligent Systems, 2017

Nowadays, component identification is one of the main challenges of software analysis and design. The component identification process aims at clustering classes into components and subcomponents. There are a number of methods to identify components in the literature; however, most of them cannot be customized to software architect’s preferences. To address this limitation, in this paper, we propose a preference-based method by the name of preference-based component identification using particle swarm optimization (PCI-PSO) to identify logical components. PCI-PSO provides a novel method to handle the software architect’s preferences using an interactive (i.e. human in the loop) search. PCI-PSO employs a customized PSO to automatically classify classes into suitable logical components and avoid the problem of identifying the proper number of components. We evaluated the effectiveness of PCI-PSO with four real-world cases. Results revealed that PCI-PSO has an ability to identify more ...

PLOS ONE, 2021

Sarcasm is the main reason behind the faulty classification of tweets. It brings a challenge in natural language processing (NLP) as it hampers the method of finding people’s actual sentiment. Various feature engineering techniques are being investigated for the automatic detection of sarcasm. However, most related techniques have always concentrated only on the content-based features in sarcastic expression, leaving the contextual information in isolation. This leads to a loss of the semantics of words in the sarcastic expression. Another drawback is the sparsity of the training data. Due to the word limit of microblog, the feature vector’s values for each sample constructed by BoW produces null features. To address the above-named problems, a Multi-feature Fusion Framework is proposed using two classification stages. The first stage classification is constructed with the lexical feature only, extracted using the BoW technique, and trained using five standard classifiers, including...

Advances in Data Mining. Applications and Theoretical Aspects, 2017

Nowadays, text classification has been extensively employed in medical domain to classify free text clinical reports. In this study, text classification techniques have been used to determine cause of death from free text forensic autopsy reports using proposed term-based and SNOMED CT concept-based features. In this study, detailed term-based features and concept-based features were extracted from a set of 1500 forensic autopsy reports belonging to four manners of death and 16 different causes of death. These features were used to train text classifier. The classifier was deployed in cascade architecture: the first level will predict the manner of death and the second level will predict the CoD using proposed term-based and SNOMED CT concept-based features. Moreover, to show the significance of our proposed approach, we compared the results of our proposed approach with four state-of-the-art feature extraction approaches. Finally, we also presented the comparison of one-level classification versus two-level classification. The experimental results showed that our proposed approach showed 8% improvement in accuracy as compared to other four baselines. Moreover, two-level classification showed improved accuracy in determining CoD compared to one-level classification.

2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), 2018

Understanding how a classifier partitions a high-dimensional input space and assigns labels to the parts is an important task in machine learning. Current methods for this task mainly use color-coded sample scatterplots, which do not explicitly show the actual decision boundaries or confusion zones. We propose an image-based technique to improve such visualizations. The method samples the 2D space of a dimensionality-reduction projection and color-code relevant classifier outputs, such as the majority class label, the confusion, and the sample density, to render a dense depiction of the high-dimensional decision boundaries. Our technique is simple to implement, handles any classifier, and has only two simple-to-control free parameters. We demonstrate our proposal on several real-world high-dimensional datasets, classifiers, and two different dimensionality reduction methods.

Multimedia Tools and Applications

In this paper we review a wide spectrum of techniques which have been proposed in literature to enable acceptable recognition of language and text by machines. We discuss many techniques which have been proposed by researchers in the field of term weighting and explore the mathematical foundations of these methods. Term weighting schemes have broadly been classified as supervised and statistical methods and we present numerous examples from both categories to highlight the difference in approaches between the two broad categories. We pay particular attention to the Vector Space Model and its variants which form the basis of many of the other methods which have been discussed in the paper.

Computational Intelligence and Neuroscience

With the rapid advancement of information technology, online information has been exponentially growing day by day, especially in the form of text documents such as news events, company reports, reviews on products, stocks-related reports, medical reports, tweets, and so on. Due to this, online monitoring and text mining has become a prominent task. During the past decade, significant efforts have been made on mining text documents using machine and deep learning models such as supervised, semisupervised, and unsupervised. Our area of the discussion covers state-of-the-art learning models for text mining or solving various challenging NLP (natural language processing) problems using the classification of texts. This paper summarizes several machine learning and deep learning algorithms used in text classification with their advantages and shortcomings. This paper would also help the readers understand various subtasks, along with old and recent literature, required during the proces...

Communication Methods and Measures, 2017

This article presents an innovative methodology to study computermediated communication (CMC), which allows analysis of the multi-layered effects of online expression and reception. The methodology is demonstrated by combining the following three data sets collected from a widely tested eHealth system, the Comprehensive Health Enhancement Support System (CHESS): (1) a flexible and precise computer-aided content analysis; (2) a record of individual message posting and reading; and (3) longitudinal survey data. Further, this article discusses how the resulting data can be applied to online social network analysis and demonstrates how to construct two distinct types of online social networks-open and targeted communication networks-for different types of content embedded in social networks. As the field of communication increasingly turns its attention to online communication, it is important to consider the ways in which we can advance our ability to study communication in these contexts. In many ways, the field has been trying to catch up with the changes occurring in communication patterns as a result of the technological advancements of digital media. Traditionally, for example, a reception-effects paradigm has dominated much of communication research, especially in the field of mass communication, which assumes most influence results from exposure to messages, as the ideas that people encounter either inform or persuade (Fishbein & Cappella, 2006; Lasswell, 1950). With the advent of computer-mediated communication (CMC), however, an expression-effects approach is emerging as a significant alternative to this dominant approach (Pingree, 2007; Shah, 2016). Indeed, an impressive body of research examines the effects of CMC expression on the message sender's psychological outcomes (

Journal of Intercultural Management

Objective: In our study, we analyze Central and Eastern European (CEE) scientific papers published in peer-reviewed scientific journals between 2015 and 2021. We examine what category systems and methods are used in Central and Eastern European start-up researches in the recent years. Methodology: Our used methodology was structured literature review analysis and artificial intelligence-based natural language processing which is one of the most evolving methodological directions in economics and social sciences at present but it is very rarely used in review analysis of startup research. Value Added: The NLP method has not been widely used for the analysis of the startup literature. Furthermore, our study is the first which analyzes CEE startups research with NLP technique. Findings: Based on our results, it can be stated that CEE startup researches follow the big global startup research narratives. However, a specific conceptual network is also emerging which contains several shift...

Intelligenza Artificiale, 2019

Communications in Computer and Information Science, 2017

Automatic layout analysis has proven to be extremely important in the process of digitization of large amounts of documents. In this paper we present a mixed approach to layout analysis, introducing a SVM-aided layout segmentation process and a classification process based on local and geometrical features. The final output of the automatic analysis algorithm is a complete and structured annotation in JSON format, containing the digitalized text as well as all the references to the illustrations of the input page, and which can be used by visualization interfaces as well as annotation interfaces. We evaluate our algorithm on a large dataset built upon the first volume of the "Enciclopedia Treccani".

Knowledge Mining

The paper presents a platform that facilitates the use of tools for collecting domain specific web pages as well as for extracting information from them. It also supports the configuration of such tools to new domains and languages. The platform provides a user friendly interface through which the user can specify the domain specific resources (ontology, lexica, corpora for the training and testing of the tools), train the collection and extraction tools using these resources, and test the tools with various configurations. The platform design is based on the methodology proposed for web information retrieval and extraction in the context of the R&D project CROSSMARC.

Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 2020

facilitated the review of lengthy issues. ArguLens provides the direction of designing valuable tools for high-level reasoning and effective discussion about usability.

Lecture Notes in Computer Science, 2011

User-generated content has dominated the web's recent growth and today the so-called real-time web provides us with unprecedented access to the real-time opinions, views, and ratings of millions of users. For example, Twitter's 200m+ users are generating in the region of 1000+ tweets per second. In this work, we propose that this data can be harnessed as a useful source of recommendation knowledge. We describe a social news service called Buzzer that is capable of adapting to the conversations that are taking place on Twitter to ranking personal RSS subscriptions. This is achieved by a content-based approach of mining trending terms from both the public Twitter timeline and from the timeline of tweets published by a user's own Twitter friend subscriptions. We also present results of a live-user evaluation which demonstrates how these ranking strategies can add better item filtering and discovery value to conventional recency-based RSS ranking techniques.

European Journal of Information Systems, 2017

Online reviews have become extremely valuable sources of information about products and their customers as electronic commerce continues to proliferate rapidly. Previous research has shown that reviews of a product change and evolve over its life. Identifying and understanding patterns of change in reviews and the forces that shape them is an underexplored topic with substantial potential for predicting and improving the market performance of products. In this study, we analyze review text of nearly 50 products over the course of their lives. Our longitudinal analysis of reviews reveals changes in certain personalityrelated characteristics of buyers in ways that are consistent with the predictions of product adoption and diffusion theories. The main findings and conclusions still hold when we replicate the same procedure on reviews of a different product category. Accordingly, based on online user-generated content in the form of online reviews, this research introduces a novel empirical method for identifying the product adoption and diffusion stage. Implications of the study for theory, methodology, and practice are discussed.

Journal of the Association for Information Science and Technology, 2014

Understanding specific patterns or knowledge of selfdisclosing health information could support public health surveillance and healthcare. This study aimed to develop an analytical framework to identify selfdisclosing health information with unusual messages on web forums by leveraging advanced text-mining techniques. To demonstrate the performance of the proposed analytical framework, we conducted an experimental study on 2 major human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS) forums in Taiwan. The experimental results show that the classification accuracy increased significantly (up to 83.83%) when using features selected by the information gain technique. The results also show the importance of adopting domain-specific features in analyzing unusual messages on web forums. This study has practical implications for the prevention and support of HIV/AIDS healthcare. For example, public health agencies can re-allocate resources and deliver services to people who need help via social media sites. In addition, individuals can also join a social media site to get better suggestions and support from each other.

Lecture Notes in Computer Science, 2009

This paper is concerned with automatic classification of texts in a medical domain. The process consists in classifying reports of medical discharges into classes defined by the CIE-9-MC codes. We will assign CIE-9-MC codes to reports using either a knn model or support vector machines. One of the added values of this work is the construction of the collection using the discharge reports of a medical service. This is a difficult collection because of the high number of classes and the uneven balance between classes. In this work we study different representations of the collection, different classication models, and different weighting schemes to assign CIE-9-MC codes. Our use of document expansion is particularly novel: the training documents are expanded with the descriptions of the assigned codes taken from CIE-9-MC. We also apply SVMs to produce a ranking of classes for each test document. This innovative use of SVM offers good results in such a complicated domain.

Handbook of Research on Pattern Engineering System Development for Big Data Analytics, 2018

E-commerce has become a daily activity in human life. In it, the opinion and past experience related to particular product of others is playing a prominent role in selecting the product from the online market. In this chapter, the authors consider Tweets as a point of source to express users' emotions on particular subjects. This is scored with different sentiment scoring techniques. Since the patterns used in social media are relatively short, exact matches are uncommon, and taking advantage of partial matches allows one to significantly improve the accuracy of analysis on sentiments. The authors also focus on applying artificial neural fuzzy inference system (ANFIS) to train the model for better opinion mining. The scored sentiments are then classified using machine learning algorithms like support vector machine (SVM), decision tree, and naive Bayes.

International Journal of Semantic Computing, 2016

The volume of documents and online resources has been increasing significantly on the Web for many years. Effectively, organizing this huge amount of information has become a challenging problem. Tagging is a mechanism to aggregate information and a great step towards the Semantic Web vision. Tagging aims to organize, summarize, share and search the Web resources in an effective way. One important problem facing tagging systems is to automatically determine the most appropriate tags for Web documents. In this paper, we propose a probabilistic topic model that incorporates DBpedia knowledge into the topic model for tagging Web pages and online documents with topics discovered in them. Our method is based on integration of the DBpedia hierarchical category network with statistical topic models, where DBpedia categories are considered as topics. We have conducted extensive experiments on two different datasets to demonstrate the effectiveness of our method.

Social Network Analysis and Mining, 2016

Medical social networks have become an exchange of opinions between patients and health professionals. However, patients are anxious to quickly find a reliable analysis and a concise explanation of their medical images and express their queries through a textual description or a visual description or both sets. For this, we present in this paper a multimodal research model to research medical images based on multimedia information that is extracted from a radiological collaborative social network. Indeed, the opinions shared on a medical image in a medico-social network are a textual description which in most cases requires cleaning by using a medical thesaurus. In addition, we describe the textual description and medical image in a TF-IDF weight vector using an approach of ''bag of words''. We use latent semantic analysis to establish relationships between textual terms and visual terms in shared opinions on the medical image. The multimodal modeling researches the medical information through multimodal queries. Our model is evaluated against the ImageCLEFMed'2015 baseline, which is the ground truth for our experiments. We have conducted numerous experiments with different descriptors and many combinations of modalities. The analysis of results shows that the model based on two methods can increase the performance of a research system based on a single modality, visual or textual.

PLOS ONE, 2020

This work presents an alternative method to represent documents based on LDA (Latent Dirichlet Allocation) and how it affects to classification algorithms, in comparison to common text representation. LDA assumes that each document deals with a set of predefined topics, which are distributions over an entire vocabulary. Our main objective is to use the probability of a document belonging to each topic to implement a new text representation model. This proposed technique is deployed as an extension of the Weka software as a new filter. To demonstrate its performance, the created filter is tested with different classifiers such as a Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), and Naive Bayes in different documental corpora (OHSUMED, Reuters-21578, 20Newsgroup, Yahoo! Answers, YELP Polarity, and TREC Genomics 2015). Then, it is compared with the Bag of Words (BoW) representation technique. Results suggest that the application of our proposed filter achieves similar accura...

Applied Intelligence, 2019

Class imbalance is one of the critical areas in classification. The challenges become more severe when the data set has a large number of features. Traditional classifiers generally favour the majority class because of skewed class distributions. In recent years, feature selection is being used to select the appropriate features for better classification of minority class. However, these studies are limited to imbalance that arise between the classes. In addition to between class imbalance, within class imbalance, along with large number of features, adds additional complexity and results in poor performance of the classifier. In the current study, we propose an effective distance based feature selection method (ED-Relief) that uses a sophisticated distance measure, in order to tackle simultaneous occurrence of between and within class imbalance. This method has been tested on a variety of simulated experiments and real life data sets and the results are compared with the traditional Relief method and some of the well known recent distance based feature selection methods. The results clearly show the superiority of the proposed effective distance based feature selection method.

Knowledge Science, Engineering and Management, 2009

Web information gathering surfers from the problems of information mismatching and overloading. In an attempt to solve these fundamental problems, many works have proposed to use concept-based techniques to perform personalized information gathering for Web users. These works have significantly improved the performance of Web information gathering systems. In this paper, a survey is conducted on these works. The reviewed scholar report that the concept-based, personalized techniques can gather more useful and meaningful information for Web users. The survey also suggests that improvement is needed for the representation and acquisition of user profiles in personalized Web information gathering.

Journal of Quantitative Linguistics, 2013

The use of naive Bayesian classifier (NB) and the classifier by the k nearest neighbors (kNN) in classification semantic analysis of authors' texts of English fiction has been analysed. The authors' works are considered in the vector space the basis of which is formed by the frequency characteristics of semantic fields of nouns and verbs. Highly precise classification of authors' texts in the vector space of semantic fields indicates about the presence of particular spheres of author's idiolect in this space which characterizes the individual author's style.

Requirements Engineering, 2012

Online feature request management systems are popular tools for gathering stakeholders' change requests during system evolution. Deciding which feature requests require attention and how much upfront analysis to perform on them is an important problem in this context: too little upfront analysis may result in inadequate functionalities being developed, costly changes, and wasted development effort; too much upfront analysis is a waste of time and resources. Early predictions about which feature requests are most likely to fail due to insufficient or inadequate upfront analysis could facilitate such decisions. Our objective is to study whether it is possible to make such predictions automatically from the characteristics of the online discussions on feature requests. This paper presents a study of feature request failures in seven large projects, an automated tool-implemented framework for constructing failure prediction models, and a comparison of the performance of the different prediction techniques for these projects. The comparison relies on a cost-benefit model for assessing the value of additional upfront analysis. In this model, the value of additional upfront analysis depends on its probability of success in preventing failures and on the relative cost of the failures it prevents compared to its own cost. We show that for reasonable estimations of these two parameters automated prediction models provide more value than a set of baselines for many failure types and projects. This suggests automated failure prediction during requirements elicitation to be a promising approach for guiding requirements engineering efforts in online settings.

Entropy, 2017

Overfitting is an important problem in machine learning. Several algorithms, such as the extreme learning machine (ELM), suffer from this issue when facing high-dimensional sparse data, e.g., in text classification. One common issue is that the extent of overfitting is not well quantified. In this paper, we propose a quantitative measure of overfitting referred to as the rate of overfitting (RO) and a novel model, named AdaBELM, to reduce the overfitting. With RO, the overfitting problem can be quantitatively measured and identified. The newly proposed model can achieve high performance on multi-class text classification. To evaluate the generalizability of the new model, we designed experiments based on three datasets, i.e., the 20 Newsgroups, Reuters-21578, and BioMed corpora, which represent balanced, unbalanced, and real application data, respectively. Experiment results demonstrate that AdaBELM can reduce overfitting and outperform classical ELM, decision tree, random forests, and AdaBoost on all three text-classification datasets; for example, it can achieve 62.2% higher accuracy than ELM. Therefore, the proposed model has a good generalizability.

Neural Computing and Applications, 2016

In this paper, we propose a novel hybrid text classification model based on deep belief network and softmax regression. To solve the sparse high-dimensional matrix computation problem of texts data, a deep belief network is introduced. After the feature extraction with DBN, softmax regression is employed to classify the text in the learned feature space. In pre-training procedures, the deep belief network and softmax regression are first trained, respectively. Then, in the fine-tuning stage, they are transformed into a coherent whole and the system parameters are optimized with Limited-memory Broyden-Fletcher-Goldfarb-Shanno algorithm. The experimental results on Reuters-21,578 and 20-Newsgroup corpus show that the proposed model can converge at fine-tuning stage and perform significantly better than the classical algorithms, such as SVM and KNN.

ArXiv, 2021

In this work, we provide a follow-up on the Moldavian versus Romanian Cross-Dialect Topic Identification (MRC) shared task of the VarDial 2019 Evaluation Campaign. The shared task included two sub-task types: one that consisted in discriminating between the Moldavian and the Romanian dialects and one that consisted in classifying documents by topic across the two dialects of Romanian. Participants achieved impressive scores, e.g. the top model for Moldavian versus Romanian dialect identification obtained a macro F1 score of 0.895. We conduct a subjective evaluation by human annotators, showing that humans attain much lower accuracy rates compared to machine learning (ML) models. Hence, it remains unclear why the methods proposed by participants attain such high accuracy rates. Our goal is to understand (i) why the proposed methods work so well (by visualizing the discriminative features) and (ii) to what extent these methods can keep their high accuracy levels, e.g. when we shorten ...

Lecture Notes in Computer Science, 2011

This paper proposes a system enforcing content-based message filtering conceived as a key service for On-line Social Networks (OSNs). The system allows OSN users to have a direct control on the messages posted on their walls. This is achieved trough a flexible rule-based system, that allows a user to customize the filtering criteria to be applied to their walls, and a Machine Learning based soft classifier automatically producing membership labels in support of content-based filtering.

Journal of the American Society for Information Science and Technology, 2010

The purpose of this study is to examine whether the understandings of subject-indexing processes conducted by human indexers have a positive impact on the effectiveness of automatic subject term assignment through text categorization (TC). More specifically, human indexers' subject-indexing approaches, or conceptions, in conjunction with semantic sources were explored in the context of a typical scientific journal article dataset. Based on the premise that subject indexing approaches or conceptions with semantic sources are important for automatic subject term assignment through TC, this study proposed an indexing conception-based framework. For the purpose of this study, two research questions were explored: To what extent are semantic sources effective? To what extent are indexing conceptions effective? The experiments were conducted using a Support Vector Machine implementation in WEKA (I.H. Witten & E. Frank, 2000). Using F-measure, the experiment results showed that cited works, source title, and title were as effective as the full text while a keyword was found more effective than the full text. In addition, the findings showed that an indexing conceptionbased framework was more effective than the full text. The content-oriented and the document-oriented indexing approaches especially were found more effective than the full text. Among three indexing conceptionbased approaches, the content-oriented approach and the document-oriented approach were more effective than the domain-oriented approach. In other words, in the context of a typical scientific journal article dataset, the objective contents and authors' intentions were more desirable for automatic subject term assignment via TC

Cluster Computing

Misleading health information is a critical phenomenon in our modern life due to advance in technology. In fact, social media facilitated the dissemination of information, and as a result, misinformation spread rapidly, cheaply, and successfully. Fake health information can have a significant effect on human behavior and attitudes. This survey presents the current works developed for misleading information detection (MLID) in health fields based on machine learning and deep learning techniques and introduces a detailed discussion of the main phases of the generic adopted approach for MLID. In addition, we highlight the benchmarking datasets and the most used metrics to evaluate the performance of MLID algorithms are discussed and finally, a deep investigation of the limitations and drawbacks of the current progressing technologies in various research directions is provided to help the researchers to use the most proper methods in this emerging task of MLID.

International Journal of Computer Processing of Languages, 2005

In this paper, we present a novel model for improving the performance of Domain Dictionary-based text categorization. The proposed model is named as Self-Partition Model (SPM). SPM can group the candidate words into the predefined clusters, which are generated according to the structure of Domain Dictionary. Using these learned clusters as features, we proposed a novel text representation. The experimental results show that the proposed text representation-based text categorization system performs better than the Domain Dictionary-based text categorization system. It also performs better than the system based on Bagof-Words when the number of features is small and the training corpus size is small.

Communications in Computer and Information Science, 2017

In this paper, we present a semi-supervised learning algorithm for classification of text documents. A method of labeling unlabeled text documents is presented. The presented method is based on the principle of divide and conquer strategy. It uses recursive K-means algorithm for partitioning both labeled and unlabeled data collection. The K-means algorithm is applied recursively on each partition till a desired level partition is achieved such that each partition contains labeled documents of a single class. Once the desired clusters are obtained, the respective cluster centroids are considered as representatives of the clusters and the nearest neighbor rule is used for classifying an unknown text document. Series of experiments have been conducted to bring out the superiority of the proposed model over other recent state of the art models on 20Newsgroups dataset.

Applied Sciences, 2021

The exponential growth of documents in the web makes it very hard for researchers to be aware of the relevant work being done within the scientific community. The task of efficiently retrieving information has therefore become an important research topic. The objective of this study is to test how the efficiency of the text classification changes if different weights are previously assigned to the sections that compose the documents. The proposal takes into account the place (section) where terms are located in the document, and each section has a weight that can be modified depending on the corpus. To carry out the study, an extended version of the OHSUMED corpus with full documents have been created. Through the use of WEKA, we compared the use of abstracts only with that of full texts, as well as the use of section weighing combinations to assess their significance in the scientific article classification process using the SMO (Sequential Minimal Optimization), the WEKA Support V...

Applied Intelligence, 2020

In recent years there has been a significant rethinking of corporate management, which is increasingly based on customer orientation principles. As a matter of fact, customer relationship management processes and systems are ever more popular and crucial to facing today’s business challenges. However, the large number of available customer communication stimuli coming from different (direct and indirect) channels, require automatic language processing techniques to help filter and qualify such stimuli, determine priorities, facilitate the routing of requests and reduce the response times. In this scenario, sentiment analysis plays an important role in measuring customer satisfaction, tracking consumer opinion, interacting with consumers and building customer loyalty. The research described in this paper proposes an approach based on Hierarchical Attention Networks for detecting the sentiment polarity of customer communications. Unlike other existing approaches, after initial trainin...

Terminology, 2015

In this paper we make two contributions. First, we describe a multi-component system called BiTES (Bilingual Term Extraction System) designed to automatically gather domain-specific bilingual term pairs from Web data. BiTES components consist of data gathering tools, domain classifiers, monolingual text extraction systems and bilingual term aligners. BiTES is readily extendable to new language pairs and has been successfully used to gather bilingual terminology for 24 language pairs, including English and all official EU languages, save Irish. Second, we describe a novel set of methods for evaluating the main components of BiTES and present the results of our evaluation for six language pairs. Results show that the BiTES approach can be used to successfully harvest quality bilingual term pairs from the Web. Our evaluation method delivers significant insights about the strengths and weaknesses of our techniques. It can be straightforwardly reused to evaluate other bilingual term extr...

Machine Learning and Data Mining in Pattern Recognition, 2012

Text Categorization (TC) has attracted the attention of the research community in the last decade. Algorithms like Support Vector Machines, Naïve Bayes or k Nearest Neighbors have been used with good performance, confirmed by several comparative studies. Recently, several ensemble classifiers were also introduced in TC. However, many of those can only provide a category for a given new sample. Instead, in this paper, we propose a methodology-MECAC-to build an ensemble of classifiers that has two advantages to other ensemble methods: 1) it can be run using parallel computing, saving processing time and 2) it can extract important statistics from the obtained clusters. It uses the mean co-association matrix to solve binary TC problems. Our experiments revealed that our framework performed, on average, 2.04% better than the best individual classifier on the tested datasets. These results were statistically validated for a significance level of 0.05 using the Friedman Test.

Arabian journal for science and engineering, 2022

With the development of websites and social networks, Internet users generate a massive amount of comments and information on the Web. Sentiment analysis, also called opinion mining, offers an opportunity to mine the people's sentiments and emotions from the textual comments. In the last decade, sentiment analysis has been applied in research areas such as recommendation and support systems and has become an area of interest for many researchers. Therefore, many studies have been carried out on English, while other languages, such as Arabic, received less attention. Increasingly, sentiment analysis researchers use machine learning due to its excellent performance. However, the generated models are black boxes and noninterpretable by the users. The rule-based classification is a promising approach for generating interpretable models. This work proposes a classification rule-based Arabic sentiment analysis approach together with a new binary equilibrium optimization metaheuristic algorithm as an optimization method for classification rule generation from Arabic documents. The proposed approach has been experimented on the Opinion Corpus for Arabic (OCA) and generates a classification model of thirteen rules. The comparison results with state-of-the-art methods show that the proposed approach outperforms all other white-box models regarding classification accuracy.

Software: Practice and Experience, 2020

Source code comments contain key information about the underlying software system. Many redocumentation approaches, however, cannot exploit this valuable source of information. This is mainly due to the fact that not all comments have the same goals and target audience and can therefore only be used selectively for redocumentation. Performing a required classification manually, for example, in the form of heuristics, is usually time-consuming and error-prone and strongly dependent on programming languages and guidelines of concrete software systems. By leveraging machine learning (ML), it should be possible to classify comments and thus transfer valuable information from the source code into documentation with less effort but the same quality. We applied classical ML techniques but also deep learning (DL) approaches to legacy systems by transferring source code comments into meaningful representations using, for example, word embeddings but also novel approaches using quick response codes or a special character-to-image encoding. The results were compared with industry-strength heuristic classification. As a result, we found that ML outperforms the heuristics in number of errors and less effort, that is, we finally achieve an accuracy of more than 95% for an image-based DL network and even over 96% for a traditional approach using a random forest classifier.

Journal of Computer Science and Technology, 2012

The large number of new bug reports received in bug repositories of software systems makes their management a challenging task. Handling these reports manually is time consuming, and often results in delaying the resolution of important bugs. To address this issue, a recommender may be developed which automatically prioritizes the new bug reports. In this paper, we propose and evaluate a classification based approach to build such a recommender. We use the Naïve Bayes and Support Vector Machine (SVM) classifiers, and present a comparison to evaluate which classifier performs better in terms of accuracy. Since a bug report contains both categorical and text features, another evaluation we perform is to determine the combination of features that better determines the priority of a bug. To evaluate the bug priority recommender, we use precision and recall measures and also propose two new measures, Nearest False Negatives (NFN) and Nearest False Positives (NFP), which provide insight into the results produced by precision and recall. Our findings are that the results of SVM are better than the Naïve Bayes algorithm for text features, whereas for categorical features, Naïve Bayes performance is better than SVM. The highest accuracy is achieved with SVM when categorical and text features are combined for training.

Multimedia Tools and Applications

Efficient word representation techniques (word embeddings) with modern machine learning models have shown reasonable improvement on automatic text classification tasks. However, the effectiveness of such techniques has not been evaluated yet in terms of insufficient word vector representation for training. Convolutional Neural Network has achieved significant results in pattern recognition, image analysis, and text classification. This study investigates the application of the CNN model on text classification problems by experimentation and analysis. We trained our classification model with a prominent word embedding generation model, Fast Text on publically available datasets, six benchmark datasets including Ag News, Amazon Full and Polarity, Yahoo Question Answer, Yelp Full, and Polarity. Furthermore, the proposed model has been tested on the Twitter US airlines non-benchmark dataset as well. The analysis indicates that using Fast Text as word embedding is a very promising approach.

Artificial Intelligence, 2021

Instagram is one of the world’s top ten most popular social networks. Instagram is the most popular social networking platform in the United States, India, and Brazil, with over 1 billion monthly active users. Each of these countries has more than 91 million Instagram users. The number of Instagram users shows the various reasons and goals for them to play this social media. Social Media Marketing does not escape being one of the purposes of using Instagram, with benefits to place a market for their products. Using text classification to categorize Instagram captions into organized groups, namely fashion, food & beverage, technology, health & beauty, lifestyle & travel, this paper is expected to help people know the current trends on Instagram. The Support Vector Machine algorithm in this research is used in 66171 post captions to classify trending on Instagram. The TF-IDF (Term Frequency times Inverse Document Frequency) method and percentage variations were used for data separatio...

Journal of Intelligent Information Systems, 2017

In the last decades, an increasing number of employers and job seekers have been relying on Web resources to get in touch and to find a job. If appropriately retrieved and analyzed, the huge number of job vacancies available today on on-line job portals can provide detailed and valuable information about the Web Labor Market dynamics and trends. In particular, this information can be useful to all actors, public and private, who play a role in the European Labor Market. This paper presents WoLMIS, a system aimed at collecting and automatically classifying multilingual Web job vacancies with respect to a standard taxonomy of occupations. The proposed system has been developed for the Cedefop European agency, which supports the development of European Vocational Education and Training

Modeling Decisions for Artificial Intelligence, 2019

Nowadays, big volumes of User-Generated Content (UGC) spread across various kinds of social media. In microblogging, UCG can be generated in the form of 'newsworthy' posts, i.e., related to information that has a public utility for the people. In this context, being the UGC diffused without almost any traditional form of trusted external control, the possibility of incurring in possible fake news is far from remote. For this reason, several approaches for fake news detection in microblogging have been proposed up to now, mostly based on machine learning techniques. In this paper, an ongoing work based on the use of the Multi-Criteria Decision Making (MCDM) paradigm to detect fake news is proposed. The aim is to reduce data dependency in building the model, and to have flexible control over the choices behind the fake news detection process.

The Scientific World Journal, 2014

This paper presents the perceptual experiments that were carried out in order to validate the methodology of transforming expressive speech styles using voice quality (VoQ) parameters modelling, along with the well-known prosody (F0, duration, and energy), from a neutral style into a number of expressive ones. The main goal was to validate the usefulness of VoQ in the enhancement of expressive synthetic speech in terms of speech quality and style identification. A harmonic plus noise model (HNM) was used to modify VoQ and prosodic parameters that were extracted from an expressive speech corpus. Perception test results indicated the improvement of obtained expressive speech styles using VoQ modelling along with prosodic characteristics.

JMIR Medical Informatics, 2020

Background The detection of infectious diseases through the analysis of free text on electronic health reports (EHRs) can provide prompt and accurate background information for the implementation of preventative measures, such as advertising and monitoring the effectiveness of vaccination campaigns. Objective The purpose of this paper is to compare machine learning techniques in their application to EHR analysis for disease detection. Methods The Pedianet database was used as a data source for a real-world scenario on the identification of cases of varicella. The models’ training and test sets were based on two different Italian regions’ (Veneto and Sicilia) data sets of 7631 patients and 1,230,355 records, and 2347 patients and 569,926 records, respectively, for whom a gold standard of varicella diagnosis was available. Elastic-net regularized generalized linear model (GLMNet), maximum entropy (MAXENT), and LogitBoost (boosting) algorithms were implemented in a supervised environme...

Mathematics

Determining the author of a text is a difficult task. Here, we compare multiple Artificial Intelligence techniques for classifying literary texts written by multiple authors by taking into account a limited number of speech parts (prepositions, adverbs, and conjunctions). We also introduce a new dataset composed of texts written in the Romanian language on which we have run the algorithms. The compared methods are artificial neural networks, multi-expression programming, k-nearest neighbour, support vector machines, and decision trees with C5.0. Numerical experiments show, first of all, that the problem is difficult, but some algorithms are able to generate acceptable error rates on the test set.

Applied Sciences

Assigning predefined classes to natural language texts, based on their content, is a necessary component in many tasks in organizations. This task is carried out by classifying documents within a set of predefined categories using models and computational methods. Text representation for classification purposes has traditionally been performed using a vector space model due to its good performance and simplicity. Moreover, the classification of texts via multilabeling has typically been approached by using simple label classification methods, which require the transformation of the problem studied to apply binary techniques, or by adapting binary algorithms. Over the previous decade, text classification has been extended using deep learning models. Compared to traditional machine learning methods, deep learning avoids rule design and feature selection by humans, and automatically provides semantically meaningful representations for text analysis. However, deep learning-based text cl...

Applied Intelligence, 2014

Recently, supervised topic modeling approaches have received considerable attention. However, the representative labeled latent Dirichlet allocation (L-LDA) method has a tendency to over-focus on the pre-assigned labels, and does not give potentially lost labels and common semantics sufficient consideration. To overcome these problems, we propose an extension of L-LDA, namely supervised labeled latent Dirichlet allocation (SL-LDA), for document categorization. Our model makes two fundamental assumptions, i.e., Prior 1 and Prior 2, that relax the restriction of label sampling and extend the concept of topics. In this paper, we develop a Gibbs expectation-maximization algorithm to learn the SL-LDA model. Quantitative experimental results demonstrate that SL-LDA is competitive with state-of-the-art approaches on both single-label and multi-label corpora.

Artificial Intelligence Review, 2016

Hadiths are important textual sources of law, tradition, and teaching in the Islamic world. Analyzing the unique linguistic features of Hadiths (e.g. ancient Arabic language and story-like text) results to compile and utilize specific natural language processing methods. In the literature, no study is solely focused on Hadith from artificial intelligence perspective, while many new developments have been overlooked and need to be highlighted. Therefore, this review analyze all academic journal and conference publications that using two main methods of artificial intelligence for Hadith text: Hadith classification and mining. All Hadith relevant methods and algorithms from the literature are discussed and analyzed in terms of functionality, simplicity, F-score and accuracy. Using various different Hadith datasets makes a direct comparison between the evaluation results impossible. Therefore, we have reimplemented and evaluated the methods using a single dataset (i.e. 3150 Hadiths from Sahih Al-Bukhari book). The result of evaluation on the classification method reveals that neural networks classify the Hadith with 94 % accuracy. This is because neural networks are capable of handling complex (high dimensional) input data. The Hadith mining method that combines vector space model, Cosine similarity, and enriched queries obtains the best accuracy result

SN Computer Science

Text classification is important to better understand online media. A major problem for creating accurate text classifiers using machine learning is small training sets due to the cost of annotating them. On this basis, we investigated how SVM and NBSVM text classifiers should be designed to achieve high accuracy and how the training sets should be sized to efficiently use annotation labor. We used a four-way repeated-measures full-factorial design of 32 design factor combinations. For each design factor combination 22 training set sizes were examined. These training sets were subsets of seven public text datasets. We study the statistical variance of accuracy estimates by randomly drawing new training sets, resulting in accuracy estimates for 98,560 different experimental runs. Our major contribution is a set of empirically evaluated guidelines for creating online media text classifiers using small training sets. We recommend uni- and bi-gram features as text representation, btc te...

Progress in Artificial Intelligence, 2019

Issue tracking systems are overall change-management tools in software development. The issue-solving life cycle is a complex socio-technical activity that requires team discussion and knowledge sharing between members. In that process, issue classification facilitates an understanding of issues and their analysis. Issue tracking systems permit the tagging of issues with default labels (e.g., bug, enhancement) or with customized team labels (e.g., test failures, performance). However, a current problem is that many issues in open-source projects remain unlabeled. The aim of this paper is to improve maintenance tasks in development teams, evaluating models that can suggest a label for an issue using its text comments. We analyze data on issues from several GitHub trending projects, first by extracting issue information and then by applying text mining classifiers (i.e., support vector machine and naive Bayes multinomial). The results suggest that very suitable classifiers may be obtained to label the issues or, at least, to suggest the most suitable candidate labels.

Expert Systems with Applications, 2022

Discovering the main features of virality patterns in Twitter is the focus of this research. Five trending topics related to the COVID-19 pandemic were selected for the study, with Spanish as the target language. To carry out the discovery of virality patterns, we applied opinion mining techniques that enable us to structure the information based on the polarity of the messages and the emotions they contain. After transforming the information from an unstructured textual representation to a structured one, data mining techniques were applied, specifically association rules mining. Message patterns with the highest virality (high shares and high likes), and at the same time the most relevant characteristics of the patterns with less impact were extracted. After an exhaustive analysis of the most relevant non-redundant rules, it can be concluded that messages with a high-negative polarity and a very high emotional charge, especially emotions that have intensified with the COVID-19 pandemic, such as fear, sadness, anger and surprise are more likely to go viral in social media. By contrast, messages with little news coverage in the media, few authors, and the absence of surprise are relevant features when it comes to seeing messages with very low dissemination in social media.

Information

Over the past few years, word embeddings and bidirectional encoder representations from transformers (BERT) models have brought better solutions to learning text representations for natural language processing (NLP) and other tasks. Many NLP applications rely on pre-trained text representations, leading to the development of a number of neural network language models for various languages. However, this is not the case for Amharic, which is known to be a morphologically complex and under-resourced language. Usable pre-trained models for automatic Amharic text processing are not available. This paper presents an investigation on the essence of learned text representation for information retrieval and NLP tasks using word embeddings and BERT language models. We explored the most commonly used methods for word embeddings, including word2vec, GloVe, and fastText, as well as the BERT model. We investigated the performance of query expansion using word embeddings. We also analyzed the use...

2021 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER)

The learning and usage of an API is supported by official documentation. Like source code, API documentation is itself a software product. Several research results show that bad design in API documentation can make the reuse of API features difficult. Indeed, similar to code smells or code antipatterns, poorly designed API documentation can also exhibit 'smells'. Such documentation smells can be described as bad documentation styles that do not necessarily produce an incorrect documentation but nevertheless make the documentation difficult to properly understand and to use. Recent research on API documentation has focused on finding content inaccuracies in API documentation and to complement API documentation with external resources (e.g., crowd-shared code examples). We are aware of no research that focused on the automatic detection of API documentation smells. This paper makes two contributions. First, we produce a catalog of five API documentation smells by consulting literature on API documentation presentation problems. We create a benchmark dataset of 1,000 API documentation units by exhaustively and manually validating the presence of the five smells in Java official API reference and instruction documentation. Second, we conduct a survey of 21 professional software developers to validate the catalog. The developers agreed that they frequently encounter all five smells in API official documentation and 95.2% of them reported that the presence of the documentation smells negatively affects their productivity. The participants wished for tool support to automatically detect and fix the smells in API official documentation. We develop a suite of rule-based, deep and shallow machine learning classifiers to automatically detect the smells. The best performing classifier BERT, a deep learning model, achieves F1-scores of 0.75-0.97.

bioRxiv (Cold Spring Harbor Laboratory), 2023

Background: The COVID-19 pandemic has led to an unprecedented amount of scientific publications, growing at a pace never seen before. Multiple living systematic reviews have been developed to assist professionals with up-to-date and trustworthy health information, but it is increasingly challenging for systematic reviewers to keep up with the evidence in electronic databases. We aimed to investigate deep learning-based machine learning algorithms to classify COVID-19 related publications to help scale-up the epidemiological curation process. Methods: In this retrospective study, five different pre-trained deep learning-based language models were fine-tuned on a dataset of 6,365 publications manually classified into two classes, three subclasses and 22 sub-subclasses relevant for epidemiological triage purposes. In a k-fold cross-validation setting, each standalone model was assessed on a classification task and compared against an ensemble, which takes the standalone model predictions as input and uses different strategies to infer the optimal article class. A ranking task was also considered, in which the model outputs a ranked list of sub-subclasses associated with the article. Results: The ensemble model significantly outperformed the standalone classifiers, achieving a F1score of 89.2 at the class level of the classification task. The difference between the standalone and ensemble models increases at the sub-subclass level, where the ensemble reaches a micro F1-score of 70% against 67% for the best performing standalone model. For the ranking task, the ensemble obtained the highest recall@3, with a performance of 89%. Using an unanimity voting rule, the ensemble can provide predictions with higher confidence .

International Journal of Information System Modeling and Design

Extracting knowledge from unstructured text and then classifying it is gaining importance after the data explosion on the web. The traditional text classification approaches are becoming ubiquitous, but the hybrid of semantic knowledge representation with statistical techniques can be more promising. The developed method attempts to fabricate neural networks to expedite and improve the simulation of ontology-based classification. This paper weighs upon the accurate results between the ontology-based text classification and traditional classification based on the artificial neural network (ANN) using distinguished parameters such as accuracy, precision, etc. The experimental analysis shows that the proposed findings are substantially better than the conventional text classification, taking the course of action into account. The authors also ran tests to compare the results of the proposed research model with one of the latest researches, resulting in a cut above accuracy and F1 score...

International Journal of Software Engineering and Knowledge Engineering, 2017

Nowadays, every manufacturer or retailer displays their product information on various websites. The customer has to visit, the number of such web pages to choose the right product, because the information is not available at one place. There are some websites that show such information in one place, but they are product specific and in general information is manually updated. In this paper, we propose a novel concept of web-spreadsheet, which displays product information by crawling through related web pages and generates information like a spreadsheet where each row represents product information and each column represents product attributes. We are extracting the product name of specified product class using decision tree-based classifier by features obtained using Part of Speech (POS) tagging and distance measure. It also extracts the value-measure pairs of preset attributes using distance measure, POS tagging and Data type. This approach will save a lot of time of comparing dif...

Computational Intelligence and Neuroscience, 2020

With the growing information on web, online movie review is becoming a significant information resource for Internet users. However, online users post thousands of movie reviews on daily basis and it is hard for them to manually summarize the reviews. Movie review mining and summarization is one of the challenging tasks in natural language processing. Therefore, an automatic approach is desirable to summarize the lengthy movie reviews, and it will allow users to quickly recognize the positive and negative aspects of a movie. This study employs a feature extraction technique called bag of words (BoW) to extract features from movie reviews and represent the reviews as a vector space model or feature vector. The next phase uses Naïve Bayes machine learning algorithm to classify the movie reviews (represented as feature vector) into positive and negative. Next, an undirected weighted graph is constructed from the pairwise semantic similarities between classified review sentences in such...

Journal of the American Society for Information Science and Technology, 2012

Detecting negative and speculative information is essential in most biomedical text-mining tasks where these language forms are used to express impressions, hypotheses, or explanations of experimental results. Our research is focused on developing a system based on machine-learning techniques that identifies negation and speculation signals and their scope in clinical texts. The proposed system works in two consecutive phases: first, a classifier decides whether each token in a sentence is a negation/speculation signal or not. Then another classifier determines, at sentence level, the tokens which are affected by the signals previously identified. The system was trained and evaluated on the clinical texts of the BioScope corpus, a freely available resource consisting of medical and biological texts: fulllength articles, scientific abstracts, and clinical reports. The results obtained by our system were compared with those of two different systems, one based on regular expressions and the other based on machine learning. Our system's results outperformed the results obtained by these two systems. In the signal detection task, the F-score value was 97.3% in negation and 94.9% in speculation. In the scope-finding task, a token was correctly classified if it had been properly identified as being inside or outside the scope of all the negation signals present in the sentence. Our proposal showed an F score of 93.2% in negation and 80.9% in speculation. Additionally, the percentage of correct scopes (those with all their tokens correctly classified) was evaluated obtaining F scores of 90.9% in negation and 71.9% in speculation.

Journal on Data Semantics, 2014

We are overwhelmed and overloaded with the data deluge brought by the digital age. Hierarchies are pervasive cognitive patterns that allow us to reorganize data and reduce the dimensionality of the information space to manageable levels (e.g., filesystems and navigational menus). In spite of their widespread adoption, such hierarchies can be improved to cope with the present needs of data sharing and reuse. First, we seldom use mechanisms to evaluate how well they partition the information space. Second, we build static and content-driven hierarchies instead of dynamic and context-driven (i.e., task-driven) ones. Third, we use ad hoc and implicit hierarchization criteria, whereas they should be explicit and shareable. This paper discusses the problems related to the construction of hierarchies, and presents a conceptual framework to turn them into reconfigurable and shareable artifacts. Moreover, it explores how dynamically reconfigurable hierarchies can better cope with the multifaceted nature of content, illustrating these principles through a tool that validates our proposal.

ISPRS International Journal of Geo-Information, 2018

This paper investigates recent research on active learning for (geo) text and image classification, with an emphasis on methods that combine visual analytics and/or deep learning. Deep learning has attracted substantial attention across many domains of science and practice, because it can find intricate patterns in big data; but successful application of the methods requires a big set of labeled data. Active learning, which has the potential to address the data labeling challenge, has already had success in geospatial applications such as trajectory classification from movement data and (geo) text and image classification. This review is intended to be particularly relevant for extension of these methods to GISience, to support work in domains such as geographic information retrieval from text and image repositories, interpretation of spatial language, and related geo-semantics challenges. Specifically, to provide a structure for leveraging recent advances, we group the relevant work into five categories: active learning, visual analytics, active learning with visual analytics, active deep learning, plus GIScience and Remote Sensing (RS) using active learning and active deep learning. Each category is exemplified by recent influential work. Based on this framing and our systematic review of key research, we then discuss some of the main challenges of integrating active learning with visual analytics and deep learning, and point out research opportunities from technical and application perspectives-for application-based opportunities, with emphasis on those that address big data with geospatial components.

Advances in Intelligent Systems and Computing, 2020

The number of news articles published on the Web has had a dramatic increase. News websites are overwhelmed daily with articles, and their processing and classification is a challenge. Reading news from the web has become an important citizen's information source, and its classification can show relevant information about social or cultural patterns on society. In this context, techniques that can automatically analyze and classify news articles are essential. In particular, data mining and machine learning techniques have been applied for the classification of web news, as they can detect structural patterns based on documents characteristics. Their use requires specialized text processing and summarizing techniques. The objective of this study is to characterize data mining and machine learning techniques used for the web news classification, the datasets used, and the evaluation metrics. We performed a systematic literature mapping of 51 primary studies published between 2000 and 2019. We found that the most used techniques fall into these paradigms: clustering, support vector machines and generative models. Also, 33 studies used online data extracted from Internet's news web pages, while 25 downloaded a previously published dataset. The most common metric is the F-measure, with 25 reports. In summary, several data mining and machine learning techniques have been applied to the automatic classification of web news, showing some trends regarding the techniques, datasets, and metrics.

Information Systems Frontiers, 2018

This paper proposes and demonstrates an approach for the often-attempted problem of market prediction, framed as classification task. We restrict our study to a widely purchased and well recognized commodity, West Texas Intermediate crude oil, which experiences significant volatility. For this purpose, nine learners using features extracted from monthly International Energy Agency (IEA) reports to predict undervalued, overvalued, and accurate valuation of the oil futures between 2003 and 2015. The often touted "Efficient Market Hypothesis" (EMH) suggests that it is impossible for individual investors to "beat the market" as market and external forces, such as geopolitical crises and natural disasters, are nearly impossible to predict. However, four algorithms were statistically better at the 95% confidence interval than "Zero-Rule" and "Random-Guess" strategies which are expected to pseudo-reflect the EMH. Furthermore, the addition of text features can significantly improve performance compared to only using price history from the oil futures data, challenging the validity of the semi-strong versions of the EMH in the crude oil market.

WSEAS TRANSACTIONS ON ENVIRONMENT AND DEVELOPMENT, 2023

The study aimed to compare the classification performance of Support Vector Machine (SVM) and Naive Bayes (NB) machine learning models for estimating customer satisfaction utilizing Filipino text. Specifically, it analyzed the characteristics of the customer satisfaction data. It also examined the impact of different model configurations, including n-gram, stop words, and stemming, on the classification performance of the two models. The research employed qualitative and quantitative methods, utilizing text analytics and sentiment analysis to extract and analyze valuable information from unstructured responses from a satisfaction survey of the University President’s leadership performance conducted among PSU personnel and students. The dataset comprised 56,000 Filipino and English-word responses, manually annotated and randomly split into training and testing datasets. The study followed a general framework encompassing data pre-processing, modeling, and model comparison. To validat...

WIRTSCHAFTSINFORMATIK, 2014

Social Media Analytics is an emerging interdisciplinary research field that aims on combining, extending, and adapting methods for analysis of social media data. On the one hand it can support IS and other research disciplines to answer their research questions and on the other hand it helps to provide architectural designs as well as solution frameworks for new social media-based applications and information systems. The authors suggest that IS should contribute to this field and help to develop and process an interdisciplinary research agenda.

Lecture Notes in Computer Science, 2016

One main challenge in modern medicine is the discovery of molecular disease subtypes characterized by relevant clinical differences, such as survival. However, clustering high-dimensional expression data is challenging due to noise and the curse of high-dimensionality. This article describes a disease subtyping pipeline that is able to exploit the important information available in pathway databases and clinical variables. The pipeline consists of a new feature selection procedure and existing clustering methods. Our procedure partitions a set of patients using the set of genes in each pathway as clustering features. To select the best features, this procedure estimates the relevance of each pathway and fuses relevant pathways. We show that our pipeline finds subtypes of patients with more distinctive survival profiles than traditional subtyping methods by analyzing a TCGA colon cancer gene expression dataset. Here we demonstrate that our pipeline improves three different clustering methods: k-means, SNF, and hierarchical clustering.

Advances in Computational Intelligence, 2017

Nowadays, Twitter depicts a rich source of on-line reviews, ratings, recommendations, and other forms of opinion expressions. This scenario has created the compelling demand to develop innovative mechanisms to store, search, organize and analyze all this data automatically. Unfortunately, it is seldom available to have enough labeled data in Twitter, because of the cost of the process or due to the impossibility to obtain them, given the rapid growing and change of this kind of media. To avoid such limitations, unsupervised categorization strategies are employed. In this paper we face the problem of cross-domain short text clustering through a compact representation that allows us to avoid the problems that arise with the high dimensionality and sparseness of vocabulary. Our experiments, conducted on a cross-domain scenario using very short texts, indicate that the proposed representation allows to generate high quality groups, according to the value of Silhouette coefficient obtained.

Wuhan University Journal of Natural Sciences, 2004

Nonnegative matrix factorization (NMF) is a method to get parts-based features of information and form the typical profiles. But the basis vectors NMF gets are not orthogonal so that parts based features of information are usually redundancy. In this paper, we propose two different approaches based on localized non negative matrix factorization (LNMF) to obtain the typical user session profiles and typical semantic profiles of junk mails. The LNMF get basis vectors as orthogonal as pos sible so that it can get accurate profiles. The experiments show that the approach based on LNMF can obtain better profiles than the approach based on NMF.

Machine Learning, 2013

Constrained clustering received a lot of attention in the last years. However, the widely used pairwise constraints are not generally applicable for hierarchical clustering, where the goal is to derive a cluster hierarchy instead of a flat partition. Therefore, we propose for the hierarchical setting-based on the ideas of pairwise constraints-the use of must-link-before (MLB) constraints. In this paper, we discuss their properties and present an algorithm that is able to create a hierarchy by considering these constraints directly. Furthermore, we propose an efficient data structure for its implementation and evaluate its effectiveness with different datasets in a text clustering scenario.

Information Retrieval, 2010

It is a big challenge to clearly identify the boundary between positive and negative streams for information filtering systems. Several attempts have used negative feedback to solve this challenge; however, there are two issues for using negative relevance feedback to improve the effectiveness of information filtering. The first one is how to select constructive negative samples in order to reduce the space of negative documents. The second issue is how to decide noisy extracted features that should be updated based on the selected negative samples. This paper proposes a pattern mining based approach to select some offenders from the negative documents, where an offender can be used to reduce the side effects of noisy features. It also classifies extracted features (i.e., terms) into three categories: positive specific terms, general terms, and negative specific terms. In this way, multiple revising strategies can be used to update extracted features. An iterative learning algorithm is also proposed to implement this approach on the RCV1 data collection, and substantial experiments show that the proposed approach achieves encouraging performance and the performance is also consistent for adaptive filtering as well. Keywords Pattern mining • Relevance feedback • Information filtering. 1 Introduction Traditional information filtering (IF) models were developed based on a term-based user profile approach (see [20,23, 15]). The advantage of term-based profiles is efficient compu

ACM Computing Surveys

Progress in natural language processing has been dictated by the rule of more : more data, more computing power, more complexity, best exemplified by deep learning Transformers. However, training (or fine-tuning) large dense models for specific applications usually requires significant amounts of computing resources. One way to ameliorate this problem is through data engineering instead of the algorithmic or hardware perspectives. Our focus here is an under-investigated data engineering technique, with enormous potential in the current scenario – Instance Selection (IS) (a.k.a. Selective Sampling, Prototype Selection). The IS goal is to reduce the training set size by removing noisy or redundant instances while maintaining or improving the effectiveness (accuracy) of the trained models and reducing the training process cost. We survey classical and recent state-of-the-art IS techniques and provide a scientifically sound comparison of IS methods applied to an essential natural langua...

Applied Sciences

Over the decades, a tremendous increase has been witnessed in the production of documents available in digital form. The increased production of documents has gained so much momentum that their rate of production jumps two-fold every five years. These articles are searched over the internet via search engines, digital libraries, and citation indexes. However, the retrieval of relevant research papers for user queries is still a pipedream. This is because scientific documents are not indexed based on some subject classification hierarchies. Hence, the classification of these documents becomes a challenging task for the researchers. Classification of the documents can be two-fold: one way is to assign a single label to each document and the other is to assign multi-labels to each document based on its belonging domains. Classification of the documents can be performed by using either the available metadata or the whole content of the documents. While performing classification, there a...

Information Security and Cryptology -- ICISC 2013, 2014

In a phishing attack, an unsuspecting victim is lured, typically via an email, to a web site designed to steal sensitive information such as bank/credit card account numbers, login information for accounts, etc. Each year Internet users lose billions of dollars to this scourge. In this paper, we present a general semantic feature selection method for text problems based on the statistical t-test and WordNet, and we show its effectiveness on phishing email detection by designing classifiers that combine semantics and statistics in analyzing the text in the email. Our feature selection method is general and useful for other applications involving text-based analysis as well. Our email body-text-only classifier achieves more than 95% accuracy on detecting phishing emails with a false positive rate of 2.24%. Due to its use of semantics, our feature selection method is robust against adaptive attacks and avoids the problem of frequent retraining needed by machine learning classifiers.

ACM Computing Surveys

The proliferation of harmful content on online platforms is a major societal problem, which comes in many different forms including hate speech, offensive language, bullying and harassment, misinformation, spam, violence, graphic content, sexual abuse, self harm, and many other. Online platforms seek to moderate such content to limit societal harm, to comply with legislation, and to create a more inclusive environment for their users. Researchers have developed different methods for automatically detecting harmful content, often focusing on specific sub-problems or on narrow communities, as what is considered harmful often depends on the platform and on the context. We argue that there is currently a dichotomy between what types of harmful content online platforms seek to curb, and what research efforts there are to automatically detect such content. We thus survey existing methods as well as content moderation policies by online platforms in this light and we suggest directions for...

Journal of Information Science, 2014

The vast number of opinions and reviews provided in Twitter is helpful in order to make interesting findings about a given industry, but given the huge number of messages published every day, it is important to detect the relevant ones. In this respect, the Twitter search functionality is not a practical tool when we want to poll messages dealing with a given set of general topics. This article presents an approach to classify Twitter messages into various topics. We tackle the problem from a linguistic angle, taking into account part-of-speech, syntactic and semantic information, showing how language processing techniques should be adapted to deal with the informal language present in Twitter messages. The TASS 2013 General corpus, a collection of tweets that has been specifically annotated to perform text analytics tasks, is used as the dataset in our evaluation framework. We carry out a wide range of experiments to determine which kinds of linguistic information have the greatest...

Creativity and Innovation Management, 2017

The notion of ‘Big Data’ has recently been attracting an increasing degree of attention from scholars and practitioners in an attempt to identify how it may be leveraged to create innovative solutions and business opportunities. Specifically, Big Data may come from a variety of sources, especially sources outside the usual boundaries of organizations, and it represents an interesting and emerging opportunity for sustaining and enhancing the effectiveness of the so‐called open innovation paradigm. However, to the best of our knowledge, no prior works have provided a broad overview of the use of Big Data for open innovation strategies. We aim to fill this gap. In particular, we have focused our investigation on two types of companies: small and medium‐sized enterprises (SMEs) and big corporations, reviewing the major academic works published so far and analysing the main industrial applications on this topic. As a result, we provide a relevant list of the main trends, opportunities, a...

Multimedia Tools and Applications, 2017

Recommending similar-interest users' groups in multimedia services is the problem of detecting for each registered user his/her membership to one interest-group of relevant consumers. The consumers in each interest-group share some relevant preferences which guarantee that the interest-group as a whole satisfies some desired properties of similarity. As a result, forming these interest-groups requires the availability of personal data of different consumers. This is a crucial requirement for different recommender systems. With the increasing trend of service providers to collect a large volume of personal data regarding their end-users, presumably to better serve them. However, a significant part of the data that is typically collected is not essential to the service being offered, or to the completion of the services it was presumably released for. Gathering such unnecessary data can be seen as a privacy threat, and storing it exposes the end-users to further unavoidable risks. In this paper, a Multimed Tools Appl

JMIR Mental Health, 2021

Background A growing body of literature has detailed the use of qualitative analyses to measure the therapeutic processes and intrinsic effectiveness of psychotherapies, which yield small databases. Nonetheless, these approaches have several limitations and machine learning algorithms are needed. Objective The objective of this study is to conduct a systematic review of the use of machine learning for automated text classification for small data sets in the fields of psychiatry, psychology, and social sciences. This review will identify available algorithms and assess if automated classification of textual entities is comparable to the classification done by human evaluators. Methods A systematic search was performed in the electronic databases of Medline, Web of Science, PsycNet (PsycINFO), and Google Scholar from their inception dates to 2021. The fields of psychiatry, psychology, and social sciences were selected as they include a vast array of textual entities in the domain of m...

Quality & Quantity

Under the influence of the health emergency triggered by the COVID-19 pandemic, many brands changed their communication strategy and included more or less explicit references to the principles of solidarity and fraternity in their TV commercials to boost the confidence and hope of Italian families during the lockdown. The traditional attitudes of the advertising format, which focused on product characteristics, were relegated to the background in order to reinforce the “brand image” through words, signs, hashtags and music that spread empathetic messages to all those who needed to regain hope and trust in a time of extreme emotional fragility. The objective of this paper is to identify the emotions and brand awareness during the lockdown using text mining techniques by measuring customer sentiment expressed on the Twitter social network. Our proposal starts from an unstructured corpus of 20,982 tweets processed with text data mining techniques to identify patterns and trends in peop...

Systematic Reviews, 2023

Background The COVID-19 pandemic has led to an unprecedented amount of scientific publications, growing at a pace never seen before. Multiple living systematic reviews have been developed to assist professionals with up-todate and trustworthy health information, but it is increasingly challenging for systematic reviewers to keep up with the evidence in electronic databases. We aimed to investigate deep learning-based machine learning algorithms to classify COVID-19-related publications to help scale up the epidemiological curation process. Methods In this retrospective study, five different pre-trained deep learning-based language models were finetuned on a dataset of 6365 publications manually classified into two classes, three subclasses, and 22 sub-subclasses relevant for epidemiological triage purposes. In a k-fold cross-validation setting, each standalone model was assessed on a classification task and compared against an ensemble, which takes the standalone model predictions as input and uses different strategies to infer the optimal article class. A ranking task was also considered, in which the model outputs a ranked list of sub-subclasses associated with the article. Results The ensemble model significantly outperformed the standalone classifiers, achieving a F1-score of 89.2 at the class level of the classification task. The difference between the standalone and ensemble models increases at the sub-subclass level, where the ensemble reaches a micro F1-score of 70% against 67% for the best-performing standalone model. For the ranking task, the ensemble obtained the highest recall@3, with a performance of 89%. Using an unanimity voting rule, the ensemble can provide predictions with higher confidence on a subset of the data, achieving detection of original papers with a F1-score up to 97% on a subset of 80% of the collection instead of 93% on the whole dataset. Conclusion This study shows the potential of using deep learning language models to perform triage of COVID-19 references efficiently and support epidemiological curation and review. The ensemble consistently and significantly outperforms any standalone model. Fine-tuning the voting strategy thresholds is an interesting alternative to annotate a subset with higher predictive confidence.

Veterinary Pathology, 2019

Machine-learning methods can assist with the medical decision-making processes at the both the clinical and diagnostic levels. In this article, we first review historical milestones and specific applications of computer-based medical decision support tools in both veterinary and human medicine. Next, we take a mechanistic look at 3 archetypal learning algorithms—naive Bayes, decision trees, and neural network—commonly used to power these medical decision support tools. Last, we focus our discussion on the data sets used to train these algorithms and examine methods for validation, data representation, transformation, and feature selection. From this review, the reader should gain some appreciation for how these decision support tools have and can be used in medicine along with insight on their inner workings.

Political Analysis, 2017

This article presents a new method and open source R package that uses syntactic information to automatically extract source-subject-predicate clauses. This improves on frequency based text analysis methods by dividing text into predicates with an identified subject and optional source, extracting the statements and actions of (political) actors as mentioned in the text. The content of these predicates can be analyzed using existing frequency based methods, allowing for the analysis of actions, issue positions and framing by different actors within a single text. We show that a small set of syntactic patterns can extract clauses and identify quotes with good accuracy, significantly outperforming a baseline system based on word order. Taking the 2008-2009 Gaza war as an example, we further show how corpus comparison and semantic network analysis applied to the results of the clause analysis can show differences in citation and framing patterns between U.S. and English-language Chinese coverage of this war.

2016

Crisis response organizations operate in very dynamic environments, in which it is essential for responders to acquire all information critical to their task execution in time. In reality, the responders are often faced with information overload, incomplete information, or a combination of both. This hampers their decisionmaking process, workflow, situational awareness and, consequently, effective execution of collaborative crisis response. Therefore, getting the right information to the right person at the right time is of crucial importance. The task of processing all data during crisis response situations and determining for whom at a particular moment the information is relevant is not straightforward. When developing an information system to support this task, some important challenges have to be taken into account. These challenges relate to the structure and truthfulness of the used data, the assessment of information relevance, and the dissemination of relevant information in time. While methods and techniques from big data can be used to collect and integrate data, machine learning can be used to build a model for relevance assessments. An example implementation of such a framework of big data is the TAID software system that collects and integrates data communicated between first responders and may send information to crisis responders that were not addressed in the initial communication. As an example of the impact of TAID on crisis response, we show its effect in a simulated crisis response scenario.

Information Systems Frontiers, 2017

Electronic Program Guides (EPGs) are systems that allow users of media applications, such as web TVs, to navigate scheduling information about current and upcoming programming. Personalized EPGs help users to overcome information overload in this domain, by exploiting recommender systems that automatically compile lists of novel and diverse video assets, based on implicitly or explicitly defined user preferences.

SN Applied Sciences, 2020

The k-nearest-neighbor (kNN) decision rule is a simple and robust classifier for text categorization. The performance of kNN decision rule depends heavily upon the value of the neighborhood parameter k. The method categorize a test document even if the difference between the number of members of two competing categories is one. Hence, choice of k is crucial as different values of k can change the result of text categorization. Moreover, text categorization is a challenging task as the text data are generally sparse and high dimensional. Note that, assigning a document to a predefined category for an arbitrary value of k may not be accurate when there is no bound on the margin of majority voting. A method is thus proposed in spirit of the nearest-neighbor decision rule using a medoid-based weighting scheme to deal with these issues. The method puts more weightage on the training documents that are not only lie close to the test document but also lie close to the medoid of its corresp...

Lecture Notes in Computer Science, 2019

Online notepad services allow users to upload and share free text anonymously. Reviewing Pastebin, one of the most popular online notepad services websites, it is possible to find textual content that could be related to illegal activities, such as leaks of personal information or hyperlinks to multimedia files containing child sexual abuse images or videos. An automatic approach to monitor and to detect these activities in such an active and a dynamic environment could be useful for Law Enforcement Agencies to fight against cybercrime. In this work, we present Pastes Content Classification 17K (PasteCC 17K), a dataset of 17640 textual samples crawled from Pastebin, which are classified in 15 categories, being 6 of them suspicious to be related to illegal ones. We used PasteCC 17K to evaluated two well-known text representation techniques, ensembled with three different supervised approaches to classify the pastes of the Pastebin website. We found that the best performance is achieved ensembling TF-IDF encoding with Logistic Regression obtaining an accuracy of 98.63%.The proposed model could assist the authorities in the detection of suspicious content shared in Pastebin.

International Studies Quarterly, 2019

New international agreements often recycle language from previous agreements, using boilerplate solutions alongside customized provisions. The presence of boilerplate in international agreements has important implications for understanding how international rules are made. The determinants behind boilerplate in international agreements have not previously been systematically evaluated. Using original data from a sample of 348 preferential trade agreements (PTAs) adopted between 1989 and 2009, we combine novel text analysis measures with Latent Order Logistic (LOLOG) graph network techniques to assess the determinants behind boilerplate in labor and environmental provisions commonly found in PTAs. Our results indicate that whereas boilerplate can be used for both efficiency and distributive purposes, international boilerplate is used primarily for efficiency gains and power-distribution considerations are not systematically important.

Applied Intelligence, 2017

This study proposes a novel scheme for termset weighting based on cardinality statistics. Specifically, termsets are evaluated by considering the number of apparent member terms. Based on a recently verified hypothesis that the occurrence of a subset of terms may also transfer worthwhile information about class memberships, the existing term weighting schemes are adapted. Here, the weight of a given termset is computed as the product of two factors. The first is a function of the member term frequencies that exist in the given document, and the second takes into account the numbers of positive and negative training documents in which the same number of members appear. By assigning a non-zero weight to the termsets when a subset of the member terms appears, the discriminative ability of different member term subsets is taken into consideration.

ACM Transactions on Software Engineering and Methodology, 2022

Sentiment analysis in software engineering (SE) has shown promise to analyze and support diverse development activities. Recently, several tools are proposed to detect sentiments in software artifacts. While the tools improve accuracy over off-the-shelf tools, recent research shows that their performance could still be unsatisfactory. A more accurate sentiment detector for SE can help reduce noise in analysis of software scenarios where sentiment analysis is required. Recently, combinations, i.e., hybrids of stand-alone classifiers are found to offer better performance than the stand-alone classifiers for fault detection. However, we are aware of no such approach for sentiment detection for software artifacts. We report the results of an empirical study that we conducted to determine the feasibility of developing an ensemble engine by combining the polarity labels of stand-alone SE-specific sentiment detectors. Our study has two phases. In the first phase, we pick five SE-specific s...

EFSA Supporting Publications, 2018

This Report presents the results from EFSA project RC/EFSA/AMU/2016/01 related to the implementation of machine learning techniques for literature reviews and systematic reviews in EFSA. An overview of the different steps of a systematic review is provided, along with possible ways for automation. Although it was found that most steps could benefit from automation, it was also observed that some steps require more sophisticated methods than those encompassed within the machine learning framework. Availability of data and methodology allowed for the development of an automatic screening tool based on several machine learning techniques. The developed shiny R application can be used for the screening of abstracts and full texts. Properties of machine learning techniques are discussed in this Report together with their most important advantages and disadvantages. The latter discussion includes both general properties, as well as context-specific properties based on their performance in three case studies. Although creating a universal automatic data extraction tool was considered to be infeasible in this stage, this step of the systematic review was addressed to allow the reviewer to scan the uploaded pdf files for certain words or string of words. Based on observations from the performed case studies, recommendations were made regarding which methods are preferred in specific situations. More explicitly, a discussion is made about the performance of the classifiers with respect to the magnitude of the pool of papers to be screened as well as to the amount of imbalance, referring to the proportion of relevant and irrelevant papers. Finally, it was concluded that the results presented in this report provide proof that the developed shiny application could be efficiently used in combination with other software such as DistillerSR.

Business & Information Systems Engineering, 2014

Social Media Analytics is an emerging interdisciplinary research field that aims on combining, extending, and adapting methods for analysis of social media data. On the one hand it can support IS and other research disciplines to answer their research questions and on the other hand it helps to provide architectural designs as well as solution frameworks for new social media-based applications and information systems. The authors suggest that IS should contribute to this field and help to develop and process an interdisciplinary research agenda.

Wiley Interdisciplinary Reviews-Data Mining and Knowledge Discovery, 2019

Health care professionals produce abundant textual information in their daily clinical practice and this information is stored in many diverse sources and, generally, in textual form. The extraction of insights from all the gathered information, mainly unstructured and lacking normalization, is one of the major challenges in computational medicine. In this respect, text mining (TM) assembles different techniques to derive valuable insights from unstructured textual data so it has led to be especially relevant in medicine. The aim of this paper is therefore to provide an extensive review of existing techniques and resources to perform TM tasks in medicine. In this review, more than 90 relevant research studies have been analyzed, describing the most important practical applications, terminological resources, tools, and open challenges of TM in medicine. This article is categorized under: Application Areas > Health Care Algorithmic Development > Biological Data Mining Algorithmic Development > Hierarchies and Trees Algorithmic Development > Model Combining K E Y W O R D S medicine, text mining, text mining tools 1 | INTRODUCTION Over the last decades, the quantity of available information daily produced in medicine is growing considerably with a special emphasis on that generated by health care professionals in their general daily practices (Feldman, Hazekamp, & Chawla, 2016). The health of the patients is regularly described by thousands of doctors; the results come in the form of textual information that is stored in different format files such as clinical records, discharge summaries, clinical monitoring sheets, or radiological reports. As a consequence of these unstructured textual data sources, the extraction of useful knowledge for decision-making and the reusability of such information is hampered. Currently, the main problem to be faced by any health care professional is not simply obtaining any available clinical information from databases but a promising subset including the most relevant and useful information. The final aim is therefore to transform this information into knowledge so professionals in the field might leverage their daily practice. Nevertheless, this is not a trivial task since clinical information is very different from any other and it usually includes some special features: high ambiguity and complex vocabulary; absence of terminological standardization; short sentences that may contain grammatical errors; overuse of acronyms; structured and unstructured data are usually combined; and texts are normally written in a narrative form. The discovery of hidden knowledge in this amount of unstructured information is essential to provide support on the decision-taking process that is carried out by the professionals every day. In this regard, the term text mining (TM) gathers the most useful techniques to derive high-quality structured information from unstructured textual data (Feldman & Sanger,

Journal of Consumer Research, 2021

This work describes and illustrates a free and easy-to-use online text-analysis tool for understanding how consumer word use varies across contexts. The tool, Wordify, uses randomized logistic regression (RLR) to identify the words that best discriminate texts drawn from different pre-classified corpora, such as posts written by men versus women, or texts containing mostly negative versus positive valence. We present illustrative examples to show how the tool can be used for such diverse purposes as (1) uncovering the distinctive vocabularies that consumers use when writing reviews on smartphones versus PCs, (2) discovering how the words used in Tweets differ between presumed supporters and opponents of a controversial ad, and (3) expanding the dictionaries of dictionary-based sentiment-measurement tools. We show empirically that Wordify's RLR algorithm performs better at discriminating vocabularies than support vector machines and chisquare selectors, while offering significant advantages in computing time. A discussion is also provided on the use of Wordify in conjunction with other textanalysis tools, such as probabilistic topic modeling and sentiment analysis, to gain more profound knowledge of the role of language in consumer behavior.

Proceedings of the 2016 ACM Symposium on Document Engineering, 2016

Taxonomies have gained a broad usage in a variety of fields due to their extensibility, as well as their use for classification and knowledge organization. Of particular interest is the digital document management domain in which their hierarchical structure can be effectively employed in order to organize documents into content-specific categories. Common or standard taxonomies (e.g., the ACM Computing Classification System) contain concepts that are too general for conceptualizing specific knowledge domains. In this paper we introduce a novel automated approach that combines sub-trees from general taxonomies with specialized seed taxonomies by using specific Natural Language Processing techniques. We provide an extensible and generalizable model for combining taxonomies in the practical context of two very large European research projects. Because the manual combination of taxonomies by domain experts is a highly time consuming task, our model measures the semantic relatedness between concept labels in CBOW or skip-gram Word2vec vector spaces. A preliminary quantitative evaluation of the resulting taxonomies is performed after applying a greedy algorithm with incremental thresholds used for matching and combining topic labels.

Communications in Computer and Information Science, 2016

To meet the fast and effective requirements of document classification in Web 2.0, the most direct strategy is to reduce the dimension of document representation without much information loss. Topic model and neural network language model are two main strategies to represent document in a low-dimensional space. To compare the effectiveness of bag-of-words, topic model and neural network language model for document classification, TF*IDF, latent Dirichlet allocation (LDA) and Paragraph Vector model are selected. Based on the generated vectors of these three methods, support vector machine classifiers are developed respectively. The performances of these three methods on English and Chinese document collections are evaluated. The experimental results show that TF*IDF outperforms LDA and Paragraph Vector, but the high-dimensional vectors take up much time and memory. Furthermore, through cross validation, the results reveal that stop words elimination and the size of training samples significantly affect the performances of LDA and Paragraph Vector, and Paragraph Vector displays its potential to overwhelm two other methods. Finally, the suggestions related with stop words elimination and data size for LDA and Paragraph Vector training are provided.

Journal of Systems Science and Systems Engineering, 2017

Societal risk classification is a fundamental and complex issue for societal risk perception. To conduct societal risk classification, Tianya Forum posts are selected as the data source, and four kinds of representations: string representation, term-frequency representation, TF-IDF representation and the distributed representation of BBS posts are applied. Using edit distance or cosine similarity as distance metric, four k-Nearest Neighbor (kNN) classifiers based on different representations are developed and compared. Owing to the priority of word order and semantic extraction of the neural network model Paragraph Vector, kNN based on the distributed representation generated by Paragraph Vector (kNN-PV) shows effectiveness for societal risk classification. Furthermore, to improve the performance of societal risk classification, through different weights, kNN-PV is combined with other three kNN classifiers as an ensemble model. Through brute force grid search method, the optimal weights are assigned to different kNN classifiers. Compared with kNN-PV, the experimental results reveal that Macro-F of the ensemble method is significantly improved for societal risk classification.

International Journal of Computational Intelligence and Applications, 2014

As more and more corporations and business entities have been publishing corporate sustainability reports, the current manual process of analyzing the reports is becoming obsolete and tedious. Development of an intelligent software tool to perform the report analysis task would be an ideal solution to this long standing problem. In this paper we argue that, given sufficient quality training using a custom corpus, corporate sustainability reports can be analyzed in mass numbers using a supervised learning based text mining software. We also discuss our methodologies of improving the accuracy of our classifier as well as the feature selector in order to gain better performance and more stability. Additionally, the achieved results of executing the developed software on one hundred reports are discussed in order to prove our claims.

Lecture Notes in Electrical Engineering, 2014

Journal of the Association for Information Science and Technology, 2021

Sentiment analysis is an emerging technique that caters for semantic orientation and opinion mining. It is increasingly used to analyse online reviews and posts for identifying people's opinions and attitudes to products and events in order to improve business performance of companies and aid to make better organising strategies of events. This paper presents an innovative approach to combining the outputs of sentiment classifiers under the framework of belief functions. It consists of the formulation of sentiment classifier outputs in the triplet evidence structure and the development of general formulas for combining triplet functions derived from sentiment classification results via three evidential combination rules along with comparative analyses. The empirical studies have been conducted on examining the effectiveness of our method for sentiment classification individually and in combination, and the results demonstrate that the best combined classifiers by our method outperforms the best individual classifiers over five review datasets.

Educational and Psychological Measurement, 2014

In this study, we explored the potential for machine scoring of short written responses to the Classroom-Video-Analysis (CVA) assessment, which is designed to measure teachers’ usable mathematics teaching knowledge. We created naïve Bayes classifiers for CVA scales assessing three different topic areas and compared computer-generated scores to those assigned by trained raters. Using cross-validation techniques, average correlations between rater- and computer-generated total scores exceeded .85 for each assessment, providing some evidence for convergent validity of machine scores. These correlations remained moderate to large when we controlled for length of response. Machine scores exhibited internal consistency, which we view as a measure of reliability. Finally, correlations between machine scores and another measure of teacher knowledge were close in size to those observed for human scores, providing further evidence for the validity of machine scores. Findings from this study s...

Springer eBooks, 2019

We briefly report on the four shared tasks organized as part of the PAN 2019 evaluation lab on digital text forensics and authorship analysis. Each task is introduced, motivated, and the results obtained are presented. Altogether, the four tasks attracted 373 registrations, yielding 72 successful submissions. This, and the fact that we continue to invite the submission of software rather than its run output using the TIRA experimentation platform, demarcates a good start into the second decade of PAN evaluations labs.

Data Science for Economics and Finance, 2021

This chapter describes the basic mechanics for building a forecasting model that uses as input sentiment indicators derived from textual data. In addition, as we focus our target of predictions on financial time series, we present a set of stylized empirical facts describing the statistical properties of lexicon-based sentiment indicators extracted from news on financial markets. Examples of these modeling methods and statistical hypothesis tests are provided on real data. The general goal is to provide guidelines for financial practitioners for the proper construction and interpretation of their own time-dependent numerical information representing public perception toward companies, stocks’ prices, and financial markets in general.

Information Systems Frontiers, 2018

Natural disasters, as well as human-made disasters, can have a deep impact on wide geographic areas, and emergency responders can benefit from the early estimation of emergency consequences. This work presents CrisMap, a Big Data crisis mapping system capable of quickly collecting and analyzing social media data. CrisMap extracts potential crisis-related actionable information from tweets by adopting a classification technique based on word embeddings and by exploiting a combination of readily-available semantic annotators to geoparse tweets. The enriched tweets are then visualized in customizable, Web-based dashboards, also leveraging ad-hoc quantitative visualizations like choropleth maps. The maps produced by our system help to estimate the impact of the emergency in its early phases, to identify areas that have been severely struck, and to acquire a greater situational awareness. We extensively benchmark the performance of our system on two Italian natural disasters by validating our maps against authoritative data. Finally, we perform a qualitative case-study on a recent devastating earthquake occurred in Central Italy.

ACM Computing Surveys, 2016

Twitter is a microblogging platform in which users can post status messages, called “tweets,” to their friends. It has provided an enormous dataset of the so-called sentiments, whose classification can take place through supervised learning. To build supervised learning models, classification algorithms require a set of representative labeled data. However, labeled data are usually difficult and expensive to obtain, which motivates the interest in semi-supervised learning. This type of learning uses unlabeled data to complement the information provided by the labeled data in the training process; therefore, it is particularly useful in applications including tweet sentiment analysis, where a huge quantity of unlabeled data is accessible. Semi-supervised learning for tweet sentiment analysis, although appealing, is relatively new. We provide a comprehensive survey of semi-supervised approaches applied to tweet classification. Such approaches consist of graph-based, wrapper-based, and...

BioMed research international, 2015

Protein interaction article classification is a text classification task in the biological domain to determine which articles describe protein-protein interactions. Since the feature space in text classification is high-dimensional, feature selection is widely used for reducing the dimensionality of features to speed up computation without sacrificing classification performance. Many existing feature selection methods are based on the statistical measure of document frequency and term frequency. One potential drawback of these methods is that they treat features separately. Hence, first we design a similarity measure between the context information to take word cooccurrences and phrase chunks around the features into account. Then we introduce the similarity of context information to the importance measure of the features to substitute the document and term frequency. Hence we propose new context similarity-based feature selection methods. Their performance is evaluated on two prote...

Social Network Analysis and Mining, 2022

The development of COVID-19 vaccines has been a great relief in many countries that have been affected by the pandemic. As a result, many governments have made significant efforts to purchase and administer vaccines to their populations. However, accommodating such vaccines is typically confronted with people's reluctance and fear. Like any other important event, COVID-19 vaccines have attracted people's discussions on social media and impacted their opinions about vaccination. Objective The goal of this study is twofold: First, it conducts a sentiment analysis around COVID-19 vaccines by automatically analyzing Arabic users' tweets. This analysis has been spread over time to better capture the changes in vaccine perceptions. This will provide us with some insights into the most popular and accepted vaccine(s) in the Arab countries, as well as the reasons behind people's reluctance to take the vaccine. Second, it develops models to detect any vaccine-related tweets, to help with gathering all information related to people's perception of the virus, and potentially detecting vaccinerelated tweets that are not necessarily tagged with the virus's main hashtags. Methods Arabic Tweets were collected by the authors, starting from January 1st, 2021, until April 20th, 2021. We deployed various Natural Language Processing (NLP) to distill our selected tweets. The curated dataset included in the analysis consisted of 1,098,376 unique tweets. To achieve the first goal, we designed state-of-the-art sentiment analysis techniques to extract knowledge related to the degree of acceptance of all existing vaccines and what are the main obstacles preventing the wide audience from accepting them. To achieve the second goal, we tackle the detection of vaccine-related tweets as a binary classification problem, where various Machine Learning (ML) models were designed to identify such tweets regardless of whether they use the vaccine hashtags or not. Results Generally, we found that the highest positive sentiments were registered for Pfizer-BioNTech, followed by Sinopharm-BIBP and Oxford-AstraZeneca. In addition, we found that 38% of the overall tweets showed negative sentiment, and only 12% had a positive sentiment. It is important to note that the majority of the sentiments vary between neutral and negative, showing the lack of conviction of the importance of vaccination among the large majority of tweeters. This paper extracts the top concerns raised by the tweets and advocates for taking them into account when advertising for the vaccination. Regarding the identification of vaccine-related tweets, the Logistic Regression model scored the highest accuracy of 0.82. Our findings are concluded with implications for public health authorities and the scholarly community to take into account to improve the vaccine's acceptance.

Applied sciences, 2021

The approach rapidly extracts reporting requirements from construction contracts and predicts overhead costs and durations associated with report preparation. Application of the approach is anticipated to provide the insights necessary to enhance contract negotiations, reporting workflow processes, and submittal procedures between clients and contractors.

ACM Transactions on Internet Technology, 2016

Argumentation mining aims at automatically extracting structured arguments from unstructured textual documents. It has recently become a hot topic also due to its potential in processing information originating from the Web, and in particular from social media, in innovative ways. Recent advances in machine learning methods promise to enable breakthrough applications to social and economic sciences, policy making, and information technology: something that only a few years ago was unthinkable. In this survey article, we introduce argumentation models and methods, review existing systems and applications, and discuss challenges and perspectives of this exciting new research area.

Information, 2019

In this paper, based on the multilingual morphological analyzer, we researched the similar low-resource languages, Uyghur and Kazakh, short text classification. Generally, the online linguistic resources of these languages are noisy. So a preprocessing is necessary and can significantly improve the accuracy. Uyghur and Kazakh are the languages with derivational morphology, in which words are coined by stems concatenated with suffixes. Usually, terms are used as the representation of text content while excluding functional parts as stop words in these languages. By extracting stems we can collect necessary terms and exclude stop words. Morpheme segmentation tool can split text into morphemes with 95% high reliability. After preparing both word- and morpheme-based training text corpora, we apply convolutional neural network (CNN) as a feature selection and text classification algorithm to perform text classification tasks. Experimental results show that the morpheme-based approach out...

Lecture Notes in Computer Science, 2015

Social messages classification is a research domain that has attracted the attention of many researchers in these last years. Indeed, the social message is different from ordinary text because it has some special characteristics like its shortness. Then the development of new approaches for the processing of the social message is now essential to make its classification more efficient. In this paper, we are mainly interested in the classification of social messages based on their spreading on online social networks (OSN). We proposed a new distance metric based on the Dynamic Time Warping distance and we use it with the probabilistic and the evidential k Nearest Neighbors (k-NN) classifiers to classify propagation networks (PrNets) of messages. The propagation network is a directed acyclic graph (DAG) that is used to record propagation traces of the message, the traversed links and their types. We tested the proposed metric with the chosen k-NN classifiers on real world propagation traces that were collected from Twitter social network and we got good classification accuracies.

Language Resources and Evaluation, 2017

This paper describes a Support Vector Machine-based system for different tasks related to the Sentiment Analysis problem in Twitter. We have paid special attention to the parameter optimization of model and to the combination of different system configurations using voting techniques. It is important to say that the tokenization and the feature selection process have been specifically designed for the Spanish Language. We have evaluated this approach in different tasks that were proposed at the five editions of the TASS workshop, from 2012 to 2016. TASS has become a framework for Sentiment Analysis tasks that is focused on the Spanish language. We describe our participation in this competition and the results achieved. We also perform an analysis and a comparison of the other participants in TASS competition. To our knowledge, the results achieved by our system are the best results published until now for the Sentiment Analysis tasks of the TASS workshop.

Knowledge and Information Systems, 2016

This paper describes a method for handling multi-class and multi-label classification problems based on the Support Vector Machine formalism. This method has been applied to the Language Identification problem in Twitter. The system evaluation was performed mainly on a Twitter dataset developed in the TweetLID workshop. This dataset contains bilingual tweets written in the most commonly used Iberian Languages (i.e., Spanish, Portuguese, Catalan, Basque, and Galician) as well as the English language. We address the following problems: 1) Social media texts. We propose a suitable tokenization that processes the peculiarities of Twitter; 2) Multilingual tweets. Since a tweet can belong to more than one language, we need to use a multi-class and multilabel classifier; 3) Similar languages. We study the main confusions among similar languages; 4) Unbalanced classes. We propose threshold-based strategy to favor classes with less data. We have also studied the use of Wikipedia and the addition of new tweets in order to increase the training dataset. Additionally, we have tested our system on Bergsma corpus, a collection of tweets in nine languages, focusing on confusable languages using the Cyrillic, Arabic, and Devanagari alphabets. To our knowledge, we obtained the best results published on the TweetLID dataset and results that are in line with the best results published on Bergsma dataset.

Agent-Mediated Knowledge Management, 2004

Distributed knowledge management systems (DKMS) have been suggested to meet the requirements of today's knowledge management. Peer-to-peer systems offer technical foundations for such distributed systems. To estimate the value of P2P-based knowledge management evaluation criteria that measure the performance of such DKMS are required. We suggest a concise framework for evaluation of such systems within different usage scenarios. Our approach is based on standard measures from the information retrieval and the databases community. These measures serve as input to a general evaluation function which is used to measure the efficiency of P2P-based KM systems. We describe test scenarios as well as the simulation software and data sets that can be used for that purpose.