Semantic Similarity Calculation

Semantic similarity measurement between words is a tedious task in web mining, information extraction and natural language processing. The semantic similarity measurement between entities is required in Web mining applications such as community extraction, identification of relations etc. In this paper, the authors proposed an automatic approach to evaluate the logical or semantic similarity between words or entities with the help of web search engines. To describe distinct word co-occurrence measures and to integrate these with lexical patterns, page counts are used. In order to identify meaningful relationships between two given words, the authors proposed a new pattern extraction algorithm and a pattern clustering algorithm. Vector Support Machine (VSM) is used to acquire the optimal combination of page counts-based co-occurrence measures and lexical pattern clusters. The proposed method overcomes various previously proposed web-based similarity measures on the benchmark data sets that showed a high correlation with human ratings.

Statistical-based collocation extraction approaches suffer from (1) low precision rate because high co-occurrence bi-grams may be syntactically unrelated and are thus not true collocations; (2) low recall rate because some true collocations with low occurrences cannot be identified successfully by statistical-based models. To integrate both syntactic rules as well as semantic knowledge into a statistical model for collocation extraction is one way to achieve a high precision while keeping a reasonable recall. This paper designs a cascade system which employs a hybrid model by integrating both syntactic and semantic knowledge into a statistical model for Chinese synonymous noun/verb collocations extraction. The grammatically bounded noun/verb collocations are extracted first from a syntactic-rule based module, which is then inputted to a semantic-based module for further retrieval of low frequent bi-gram collocations.

Recommender System is a subclass of information filtering system. It identifies similarity among users or items. It can be used as information filtering tool in online social network. Collaborative filtering recommendations are based on similarity of users or items, all data should be compared with each other in order to calculate this similarity. Due to large amount of data in dataset, too much time is required for this calculation, and in these systems, scalability problem is observed. It is better to group data, and each data should be compared with data in the same group. Content based filtering recommends items to users according to users history and items he liked in the past. To improve the working and quality of recommender system, a Hybrid approach by combining content based filtering and collaborative filtering, which includes Memory based (K-Nearest Neighbor) and Model based (clustering and rule based techniques),is presented in our proposed methodology. Using Hybrid approach, we get advantages from each other while drawbacks of both methods won't take into account.

Semantic similarity measurement between words is a tedious task in web mining, information extraction and natural language processing. The semantic similarity measurement between entities is required in Web mining applications such as community extraction, identification of relations etc. In this paper, the authors proposed an automatic approach to evaluate the logical or semantic similarity between words or entities with the help of web search engines. To describe distinct word co-occurrence measures and to integrate these with lexical patterns, page counts are used. In order to identify meaningful relationships between two given words, the authors proposed a new pattern extraction algorithm and a pattern clustering algorithm. Vector Support Machine (VSM) is used to acquire the optimal combination of page counts-based co-occurrence measures and lexical pattern clusters. The proposed method overcomes various previously proposed web-based similarity measures on the benchmark data sets that showed a high correlation with human ratings.

We describe the way to get benefit from broad cultural trends through the quantitative analysis of a vast digital book collection representing the digested history of humanity. Our research work has revealed that appropriately comparing the occurrence patterns of words in some periods of human literature can help us to accurately determine the semantic similarity between these words by means of computers without requiring human intervention. Preliminary results seem to be promising.

This paper identifies the factors that have an impact on mobile recommender systems. Recommender systems have become a technology that has been widely used by various online applications in situations where there is an information overload problem. Numerous applications such as e-Commerce, video platforms and social networks provide personalized recommendations to their users and this has improved the user experience and vendor revenues. The development of recommender systems has been focused mostly on the proposal of new algorithms that provide more accurate recommendations. However, the use of mobile devices and the rapid growth of the internet and networking infrastructure has brought the necessity of using mobile recommender systems. The links between web and mobile recommender systems are described along with how the recommendations in mobile environments can be improved. This work is focused on identifying the links between web and mobile recommender systems and to provide solid future directions that aim to lead in a more integrated mobile recommendation domain.

Semantic relation is an important concept of information science. Now a days it is widely used in semantic web. This paper aims to present a measure to automatically determine semantic relation between words using web as knowledge source. It explores whether two words are related or not even if they are dissimilar in meaning. The proposed formula is a function of frequency of occurrences of the words in each document in the corpus. This relationship measure will be useful to extract semantic information from the web. Experimental evaluation on ten manually selected word pairs using the WebKb data as information source demonstrates the effectiveness of the proposed semantic relation.

Semantic relation is an important concept of information science. Now a days it is widely used in semantic web. This paper aims to present a measure to automatically determine semantic relation between words using web as knowledge source. It explores whether two words are related or not even if they are dissimilar in meaning. The proposed formula is a function of frequency of occurrences of the words in each document in the corpus. This relationship measure will be useful to extract semantic information from the web. Experimental evaluation on ten manually selected word pairs using the WebKb data as information source demonstrates the effectiveness of the proposed semantic relation.

Despite challenges like concept drifts, or temporal dynamics in RS, RS has grown in popularity due to its usefulness in meeting customers' needs by helping them find things they might like based on past purchases and interests. Despite their great effectiveness in generating recommendations, conventional RS techniques fall short when it comes to providing accurate ideas due to problems with concept drift. The development of temporal models to account for concept drifts and guarantee more accurate recommendations has been the focus of a lot of research in the wake of these issues, giving rise to dynamic recommender systems (DRSs). However, the bulk of the effort needed to address the drift of interest is put in developing long-term and short-term models for users. It is not possible to dynamically track users' changing interests. We express doubts in this position paper about the practice of computing evaluation metrics for recommender systems as single numbers since these values only represent average effectiveness over a very long time period (for example, a year or longer). This approach only provides a vague, unchanging picture of the facts. To better understand the performance of recommender systems, we propose that researchers compute metrics across time series such as weeks or months and present the results visually, using a line chart, for instance. Insightful forecasts about an algorithm's future performance can be made with the use of results that show how an algorithm's effectiveness evolves over time. As a result, we'll be able to make more informed selections about which algorithms to utilize in a recommender system by collecting more data on an algorithm's performance over time, spotting trends, and developing more accurate forecasts about an algorithm's future performance.

Despite challenges like concept drifts, or temporal dynamics in RS, RS has grown in popularity due to its usefulness in meeting customers' needs by helping them find things they might like based on past purchases and interests. Despite their great effectiveness in generating recommendations, conventional RS techniques fall short when it comes to providing accurate ideas due to problems with concept drift. The development of temporal models to account for concept drifts and guarantee more accurate recommendations has been the focus of a lot of research in the wake of these issues, giving rise to dynamic recommender systems (DRSs). However, the bulk of the effort needed to address the drift of interest is put in developing long-term and short-term models for users. It is not possible to dynamically track users' changing interests. We express doubts in this position paper about the practice of computing evaluation metrics for recommender systems as single numbers since these values only represent average effectiveness over a very long time period (for example, a year or longer). This approach only provides a vague, unchanging picture of the facts. To better understand the performance of recommender systems, we propose that researchers compute metrics across time series such as weeks or months and present the results visually, using a line chart, for instance. Insightful forecasts about an algorithm's future performance can be made with the use of results that show how an algorithm's effectiveness evolves over time. As a result, we'll be able to make more informed selections about which algorithms to utilize in a recommender system by collecting more data on an algorithm's performance over time, spotting trends, and developing more accurate forecasts about an algorithm's future performance.

Measuring similarity between words using a search engine based on page counts alone is a challenging task. Search engines consider a document as a bag of words, ignoring the position of words in a document. In order to measure semantic similarity between two given words, this paper proposes a transformation function for web measures along with a new approach that exploits the document's title attribute and uses page counts alone returned by Web search engines. Experimental results on benchmark datasets show that the proposed approach outperforms snippets alone methods, achieving a correlation coefficient up to 71%.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Highlights  We propose a new neighbors finding method for CF based on subspace clustering.  Based on subspaces, tree structures of neighbor users are drawn for the target user.  Proposed method finds the best neighbors without any adjustable parameters.  A new similarity method is proposed to compute the similarity value.  Proposed method has been tested by Movielens 100K, Movielens 1M and Jester datasets.

Semantic similarity is a central concept that extends across numerous fields such as artificial intelligence, natural language processing, cognitive science and psychology. Accurate measurement of semantic similarity between words is essential for various tasks such as, document clustering, information retrieval, and synonym extraction. We propose a novel model of semantic similarity using the semantic relations that exist among words. Given two words, first, we represent the semantic relations that hold between those words using automatically extracted lexical pattern clusters. Next, the semantic similarity between the two words is computed using a Mahalanobis distance measure. We compare the proposed similarity measure against previously proposed semantic similarity measures on Miller-Charles benchmark dataset and WordSimilarity-353 collection. The proposed method outperforms all existing web-based semantic similarity measures, achieving a Pearson correlation coefficient of 0.867 on the Millet-Charles dataset.

Semantic similarity is a central concept that extends across numerous fields such as artificial intelligence, natural language processing, cognitive science and psychology. Accurate measurement of semantic similarity between words is essential for various tasks such as, document clustering, information retrieval, and synonym extraction. We propose a novel model of semantic similarity using the semantic relations that exist among words. Given two words, first, we represent the semantic relations that hold between those words using automatically extracted lexical pattern clusters. Next, the semantic similarity between the two words is computed using a Mahalanobis distance measure. We compare the proposed similarity measure against previously proposed semantic similarity measures on Miller-Charles benchmark dataset and WordSimilarity-353 collection. The proposed method outperforms all existing web-based semantic similarity measures, achieving a Pearson correlation coefficient of 0.867 on the Millet-Charles dataset.

Semantic similarity measurement between words is a tedious task in web mining, information extraction and natural language processing. The semantic similarity measurement between entities is required in Web mining applications such as community extraction, identification of relations etc. In this paper, the authors proposed an automatic approach to evaluate the logical or semantic similarity between words or entities with the help of web search engines. To describe distinct word co-occurrence measures and to integrate these with lexical patterns, page counts are used. In order to identify meaningful relationships between two given words, the authors proposed a new pattern extraction algorithm and a pattern clustering algorithm. Vector Support Machine (VSM) is used to acquire the optimal combination of page counts-based co-occurrence measures and lexical pattern clusters. The proposed method overcomes various previously proposed web-based similarity measures on the benchmark data sets that showed a high correlation with human ratings.

Semantic similarity measurement between words is a tedious task in web mining, information extraction and natural language processing. The semantic similarity measurement between entities is required in Web mining applications such as community extraction, identification of relations etc. In this paper, the authors proposed an automatic approach to evaluate the logical or semantic similarity between words or entities with the help of web search engines. To describe distinct word co-occurrence measures and to integrate these with lexical patterns, page counts are used. In order to identify meaningful relationships between two given words, the authors proposed a new pattern extraction algorithm and a pattern clustering algorithm. Vector Support Machine (VSM) is used to acquire the optimal combination of page counts-based co-occurrence measures and lexical pattern clusters. The proposed method overcomes various previously proposed web-based similarity measures on the benchmark data sets that showed a high correlation with human ratings.

The Matrix-Factorization (MF) based models have become popular when building Collaborative Filtering (CF) recommender systems, due to the high accuracy and scalability. Most of nowadays matrix factorization models don't have acceptable execution time during to large datasets. In this article, we introduce a new collaborative filtering recommender system, based on matrix factorization by using genetic algorithm.

Computing the textual similarity between terms (or short text expressions) that have the same meaning but which are not lexicographically similar is a key challenge in many computer related fields. The problem is that traditional approaches to semantic similarity measurement are not suitable for all situations, for example, many of them often fail to deal with terms not covered by synonym dictionaries or are not able to cope with acronyms, abbreviations, buzzwords, brand names, proper nouns, and so on. In this paper, we present and evaluate a collection of emerging techniques developed to avoid this problem. These techniques use some kinds of web intelligence to determine the degree of similarity between text expressions. These techniques implement a variety of paradigms including the study of co-occurrence, text snippet comparison, frequent pattern finding, or search log analysis. The goal is to substitute the traditional techniques where necessary.

Collaborative filtering is one of the most used approaches for providing recommendations in various online environments. Even though collaborative recommendation methods have been widely utilized due to their simplicity and ease of use, accuracy is still an issue. In this paper we propose a multi-level recommendation method with its main purpose being to assist users in decision making by providing recommendations of better quality. The proposed method can be applied in different online domains that use collaborative recommender systems, thus improving the overall user experience. The efficiency of the proposed method is shown by providing an extensive experimental evaluation using five real datasets and with comparisons to alternatives.

Extracting a subset of a given OWL ontology that captures all the ontology's knowledge about a specified set of terms is a wellunderstood task. This task can be based, for instance, on locality-based modules (LBMs). These come in two flavours, syntactic and semantic, and a syntactic LBM is known to contain the corresponding semantic LBM. For syntactic LBMs, polynomial extraction algorithms are known, implemented in the OWL API, and being used. In contrast, extracting semantic LBMs involves reasoning, which is intractable for OWL 2 DL, and these algorithms had not been implemented yet for expressive ontology languages. We present the first implementation of semantic LBMs and report on experiments that compare them with syntactic LBMs extracted from real-life ontologies. Our study reveals whether semantic LBMs are worth the additional extraction effort, compared with syntactic LBMs.

Description logics based languages have became the standard representation scheme for ontologies. They formalize the domain knowledge using interrelated concepts, contained in terminologies. The manual definition of terminologies is an expensive and error prone task, therefore automatic learning methods are a necessity. In this paper we lay the foundations of a multiple concept learning method that uses virtual concepts to aid the learning process, yielding more compact and readable terminologies. In this paper, we define virtual concepts and how they can be implemented in the current concept learning methods. We show through experiments how the method stacks up against other multiple concept learning methods.

Recommender systems are typically provided as Web 2.0 services and are part of the range of applications that give support to large-scale social networks, enabling on-line recommendations to be made based on the use of networked databases. The operating core of recommender systems is based on the collaborative filtering stage, which, in current user to user recommender processes, usually uses the Pearson correlation metric. In this paper, we present a new metric which combines the numerical information of the votes with independent information from those values, based on the proportions of the common and uncommon votes between each pair of users. Likewise, we define the reasoning and experiments on which the design of the metric is based and the restriction of being applied to recommender systems where the possible range of votes is not greater than 5. In order to demonstrate the superior nature of the proposed metric, we provide the comparative results of a set of experiments based on the MovieLens, FilmAffinity and NetFlix databases. In addition to the traditional levels of accuracy, results are also provided on the metrics' coverage, the percentage of hits obtained and the precision/recall.

Recommender systems are the efficient and most used tools that prevail over the information overload problem, provide users with the most appropriate content by considering their personal preferences (mostly, ratings). In addition to these preferences, taking into account the interaction context of users will improve the relevancy of the recommendation process. However, only a few prior studies have tried to adopt context-awareness to the recommendation model. Although a number of studies have developed recommendation models using collaborative filtering (CF), few of them have tried to adopt both CF and other artificial intelligence techniques, such as genetic algorithm (GA), as a tool to improve recommendation results. In this paper, we propose a new recommendation model, which we termed Context-Aware Collaborative Filtering using genetic algorithm (CACF-GA), for location-based advertising (LBA) based on both user's preferences and interaction's context. We first defined discrete contexts, and then applied the concept of ''context similarity'' to conventional CF to create the context-aware recommendation model. The context similarity between two contexts is designed to be optimized using GA. We collect real-world data from mobile users, build a LBA recommendation model using CACF-GA, and then perform an empirical test to validate the usefulness of CACF-GA. Experiments show our proposed model provides the most accurate prediction results compared to comparative ones.

Clustering is one of the successful approaches of the model-based collaborative filtering techniques that deals with the problem of sparsity and provides quality recommendations. In the proposed work, fuzzy c-means clustering technique is adopted in order to produce item-based clusters as well as user-based clusters. Subsequently, collaborative filtering technique explores the item-based and user-based clusters and generates the list of item-based and user-based predictions, respectively. Further, to enhance the quality of recommendations, a novel weighted hybrid scheme is designed which integrates the user-based and item-based predictions to capture the influence of each active user towards item-based and user-based predictions. The proposed schemes are further categorised on the basis of re-clustering and without re-clustering under different similarity measures over sparse and dense datasets. The experimental results reveal that the variants of the proposed hybrid schemes consistently generate better results in comparison with the corresponding variants of proposed user-based schemes and the traditional item-based schemes.

Estimating word relatedness is essential in natural language processing (NLP), and in many other related areas. Corpus-based word relatedness has its advantages over knowledge-based supervised measures. There are many corpus-based measures in the literature that can not be compared to each other as they use a different corpus. The purpose of this paper is to show how to evaluate different corpus-based measures of word relatedness by calculating them over a common corpus (i.e., the Google n-grams) and then assessing their performance with respect to gold standard relatedness datasets. We evaluate six of these measures as a starting point, all of which are re-implemented using the Google n-gram corpus as their only resource, by comparing their performance in five different data sets. We also show how a word relatedness measure based on a web search engine can be implemented using the Google n-gram corpus.

Computing the textual similarity between terms (or short text expressions) that have the same meaning but which are not lexicographically similar is a key challenge in many computer related fields. The problem is that traditional approaches to semantic similarity measurement are not suitable for all situations, for example, many of them often fail to deal with terms not covered by synonym dictionaries or are not able to cope with acronyms, abbreviations, buzzwords, brand names, proper nouns, and so on. In this paper, we present and evaluate a collection of emerging techniques developed to avoid this problem. These techniques use some kinds of web intelligence to determine the degree of similarity between text expressions. These techniques implement a variety of paradigms including the study of co-occurrence, text snippet comparison, frequent pattern finding, or search log analysis. The goal is to substitute the traditional techniques where necessary.

Description logics based languages have became the standard representation scheme for ontologies. They formalize the domain knowledge using interrelated concepts, contained in terminologies. The manual definition of terminologies is an expensive and error prone task, therefore automatic learning methods are a necessity. In this paper we lay the foundations of a multiple concept learning method that uses virtual concepts to aid the learning process, yielding more compact and readable terminologies. In this paper, we define virtual concepts and how they can be implemented in the current concept learning methods. We show through experiments how the method stacks up against other multiple concept learning methods.

The information available in the web is increasing daily. Searching for anything from web is very difficult because of availability of huge data and the disadvantage with the searching is, it simply mines data based on the keyword given by the user. It doesn’t consider the context. The only solution for searching problem is Recommendation Systems. From the past 16 years Recommendation systems changed the way of searching the products. Products are like movies, videos, research articles, news, hotels, tours, music etc. Recommendation systems are being used in each field like Ecommerce or M-commerce, social networking, research articles, music and travel. Collaborative filtering, content based filtering, Hybrid approaches and popularity based approaches are used in Recommendation Systems. Collaborative filtering mines data based on the taste and interest of similar users. Collaborative filtering implemented on neighbourhood algorithms. Content based filtering mines data based on simil...

Collaborative filtering algorithm (CF) is a personalized recommendation algorithm that is the most widely used in e-commerce. CF still needs to be improved so that it can make adequate recommendations and solve the problems such as scalability, smoothing the rating estimation. In this paper, we provide an approach of an item based collaborative filtering using item clustering prediction and including a new enhanced correlation similarity. Firstly, we cluster the items in some groups. Then, in the process of collaborative filtering recommendation, we need to calculate the similarity between the targeted item and items in the selected center. For this aim, an enhanced similarity measure based on multi criteria is proposed instead of the similarity based just on ratings' items. The objective is to consider when we calculate similarity, the integration of item rating information, the background of the item and the time-weight as criteria of the item evaluation into a convex model. In so doing, the amelioration of the similarity between items performs the recommendation. This proposed CF algorithm is showing to reduce also the influence of the former evaluation of the item.

The rapid growth of Internet technologies and availability of web tools created an opportunity to develop a robust and user-friendly web service model for medical care, and it demands urgent solutions as the uncertainty of disease spread threaten humanity. With changing Quality of Service principles, many existing web services need to offer specific medical

Semantic Similarity measures between words plays an important role in information retrieval, natural language processing and in various tasks on the web. In this paper, we have proposed a Modified Pattern Extraction Algorithm to compute the supervised semantic similarity measure between the words by combining both page count method and web snippets method. Four association measures are used to find semantic similarity between words in page count method using web search engines. We use a Sequential Minimal Optimization (SMO) support vector machines (SVM) to find the optimal combination of page counts-based similarity scores and top-ranking patterns from the web snippets method. The SVM is trained to classify synonymous word-pairs and non-synonymous word-pairs. The proposed Modified Pattern Extraction Algorithm outperforms by 89.8 percent of correlation value.

The proliferation of Internet has made people to rely on virtual recommendations. Recommender systems help out in giving important recommendations. Collaborative filtering is the most successful and widely used approach in designing recommender systems since the introduction of the concept of recommender systems. This approach uses the known tastes and preferences of a set of users to make predictions or generate recommendations about the unknown tastes and preferences of the target user. This paper discusses various works which use collaborative filtering approach to design recommender systems. The paper also gives a comparison of these approaches. IndexTerms – Recommender Systems, Collaborative filtering, Social networks. ________________________________________________________________________________________________________

We describe the way to get benefit from broad cultural trends through the quantitative analysis of a vast digital book collection representing the digested history of humanity. Our research work has revealed that appropriately comparing the occurrence patterns of words in some periods of human literature can help us to accurately determine the semantic similarity between these words by means of computers without requiring human intervention. Preliminary results seem to be promising.

Recommendation System is a subclass of information filtering system. It identifies similarity among users or items. It can be used as information filtering tool in online social network. Collaborative filtering recommendations are based on similarity of users or items, all data should be compared with each other in order to calculate this similarity. Due to large amount of data in dataset, too much time is required for this calculation, and in these systems, scalability problem is observed. It is better to group data, and each data should be compared with data in the same group. Content based filtering recommends items to users according to users history and items he liked in the past.

Semantic similarity between words is fundamental to various fields such as Cognitive Science, Artificial Intelligence, Natural Language Processing and Information Retrieval. According to Baeza-Yates and Neto [2] an Information Retrieval system “should provide the user with easy access to the information in which he is interested”. Therefore, in this domain, relying on a robust semantic similarity measure is crucial for automatic query suggestion and expansion process. In this same context, we propose a method that uses on one hand, an online English dictionary provided by the Semantic Atlas project of the French National Centre for Scientific Research (CNRS) and on the other hand, a page counts based metric returned by a social website.

Recommendation System is a subclass of information filtering system. It identifies similarity among users or items. It can be used as information filtering tool in online social network. Collaborative filtering recommendations are based on similarity of users or items, all data should be compared with each other in order to calculate this similarity. Due to large amount of data in dataset, too much time is required for this calculation, and in these systems, scalability problem is observed. It is better to group data, and each data should be compared with data in the same group. Content based filtering recommends items to users according to users history and items he liked in the past.

Semantic similarity plays a significant role in the areas of Web mining, Information Retrieval, NLP and Text mining. Even though it is exploited in various applications accurately measuring semantic similarity still remains a challenging task. In this paper a method is proposed to measure semantic similarity between words using web as information source and by combining two existing approaches to measure semantic similarity they are: Semantic Word Distance (SWD) and Snippets. The SWD measure finds the semantic similarity by determining the frequency of occurrences of the words in web pages (corpus).The semantic relation between words are also obtained through lexical patterns which are extracted from text snippets. A robust method is used to integrate these similarity scores using support vector machine. For the experimental purpose 100 word pairs are used to train the support vector machine and it classifies the word pair as either synonymous or non synonymous with higher accuracy.

Semantic similarity plays a significant role in the areas of Web mining, Information Retrieval, NLP and Text mining. Even though it is exploited in various applications accurately measuring semantic similarity still remains a challenging task. In this paper a method is proposed to measure semantic similarity between words using web as information source and by combining two existing approaches to measure semantic similarity they are: Semantic Word Distance (SWD) and Snippets. The SWD measure finds the semantic similarity by determining the frequency of occurrences of the words in web pages (corpus). The semantic relation between words are also obtained through lexical patterns which are extracted from text snippets. A

Description logics based languages have became the standard representation scheme for ontologies. They formalize the domain knowledge using interrelated concepts, contained in terminologies. The manual definition of terminologies is an expensive and error prone task, therefore automatic learning methods are a necessity. In this paper we lay the foundations of a multiple concept learning method that uses virtual concepts to aid the learning process, yielding more compact and readable terminologies. In this paper, we define virtual concepts and how they can be implemented in the current concept learning methods. We show through experiments how the method stacks up against other multiple concept learning methods.

Recommender Systems (RSs) are software tools and techniques that are used to produce recommendations for the users of a certain application in such a way that the recommendations generated are likely to be liked by the users. Popular examples of applications that use RSs include Amazon, Netflix, Spotify, and Youtube. To support the wide-spread use of RSs, a variety of open-source tool-kits have been developed. While research studies show that different algorithms work well for different recommendation scenarios and with varying data-set characteristics, it has also been pointed out that the same recommendation algorithms implemented from different tool-kits can produce significantly different results. Recommendation algorithms typically have hyper-parameters that can be used to change their behaviour. Tuning the hyper-parameters according to the recommendation scenario can effectively improve the performance of any such algorithm. Even so, RS tool-kits generally lack hyper-parameter...

I hereby certify that the information contained in this (my submission) is information pertaining to research I conducted for this project. All information other than my own contribution will be fully referenced and listed in the relevant bibliography section at the rear of the project. ALL internet material must be referenced in the bibliography section. Students are required to use the Referencing Standard specified in the report template. To use other author's written or electronic work is illegal (plagiarism) and may result in disciplinary action.

Over the past few decades, various recommendation system paradigms have been developed for both research and industrial purposes to satisfy the needs and preferences of users when they deal with enormous data. The collaborative filtering (CF) is one of the most popular recommendation techniques, although it is still immature and suffers from some difficulties such asparsity, gray sheep and scalability impeding recommendation quality. Therefore, we propose a new CF approach to deal with the gray sheep problem in order to improve the predictions accuracy. To realize this goal, our solution aims to infer new users from real ones existing in datasets. This transformation allows for creating users with opposite preferences to the real ones. On the one hand, our approach permits to amplify the number of neighbors, especially in the case of users who have unusual behavior (gray sheep). On the other hand, it facilitates building a dense similar neighborhood. The basic assumption behind this...

Finding relevant scholarly papers is an important task for researchers. Such a literature search involves identifying drawbacks in existing works and proposing new approaches that address them. However, the growing number of scientific published papers results in information overload even for simple searches, such that researchers have difficulty in finding papers relevant to their interests. Recommendation systems can help address this problem to find relevant papers efficiently. In this article, we summarize our work on scholarly paper recommendation from both relevance and serendipitous perspectives. Experimental results on a publicly-available scholarly paper recommendation dataset show that our proposed approaches provides promising recommendations for researchers, outperforming the state-of-the-art with statistical significance.

Finding relevant scholarly papers is an important task for researchers. Such a literature search involves identifying drawbacks in existing works and proposing new approaches that address them. However, the growing number of scientific published papers results in information overload even for simple searches, such that researchers have difficulty in finding papers relevant to their interests. Recommendation systems can help address this problem to find relevant papers efficiently. In this article, we summarize our work on scholarly paper recommendation from both a relevance and serendipitous perspectives. Experimental results on a publicly-available scholarly paper recommendation dataset show that our proposed approaches provide promising recommendations for researchers, outperforming the state-of-the-art with statistical significance.

Semantic similarity measures play vital roles in information retrieval, natural language processing and paraphrasing detection. With the growing plagiarisms cases in both commercial and research community, designing efficient tools and approaches for paraphrasing detection becomes crucial. This paper contrasts web-based approach related to analysis of snippets of the search engine with WordNet based measure. Several refinements of the web-based approach will be investigated and compared. Evaluations of the approaches with respect to Microsoft paraphrasing dataset will be performed and discussed.

In this paper, we propose two exponential similarity measures for collaborative filtering in recommender systems. The proposed similarity measures are used to estimate the distance between two users or items. Furthermore, an algorithm is proposed to use the distance obtained from similarity measures introduced in this paper to evaluate empty rating elements in the rating matrix and compare the results of these similarity measures with our exponential similarity measures. Therefore, we deduce that the algorithm using the exponential similarity measures works better than the algorithm using Manhattan, Euclidean, Jaccard, Cosine or Pearson similarity measures when taking K-Most-Similar-Users to evaluate the similarity in order to find ratings for the user that we want to predict ratings for.

In this paper, we build a hybrid Web-based metric for computing semantic relatedness between words. The method exploits page counts, titles, snippets and URLs returned by a Web search engine. Our technique uses traditional information retrieval methods and is enhanced by page-count-based similarity scores which are integrated with automatically extracted lexico-synantic patterns from titles, snippets and URLs for all kinds of semantically related words provided by WordNet (synonyms, hypernyms, meronyms, antonyms). A support vector machine is used to solve the arising regression problem of word relatedness and the proposed method is evaluated on standard benchmark datasets. The method achieves an overall correlation of 0.88, which is the highest among other metrics up to date.