Burst Detection

It is commonly agreed that one needs to use a threshold value in the detection of muscle activity timing in electromyographic (EMG) signal analysis. However, the algorithm for threshold determination lacks an agreement between the investigators. In this study we aimed to determine a proper threshold value in an incremental cycling exercise for accurate EMG signal analysis. Nine healthy recreationally active male subjects cycled until exhaustion. EMG recordings were performed on four low extremity muscle groups; gastrocnemius lateralis (GL), gastrocnemius medialis (GM), soleus (SOL) and vastus medialis (VM). We have analyzed our data using three different threshold levels: 25%, 35% and 45% of the mean RMS EMG value. We compared the appropriateness of these threshold values using two criteria: (1) significant correlation between the actual and estimated number of bursts and (2) proximity of the regression line of the actual and estimated number of bursts to the line of identity. It had been possible to find a significant correlation between the actual and estimated number of bursts with the 25, 35 and 45% threshold values for the GL muscle. Correlation analyses for the VM muscle had shown that the number of bursts estimated with the 35% threshold value was found to be significantly correlated with the actual number of bursts. For the GM muscle, it had been possible to predict the burst number by using either the 35% or 45% threshold value and for the SOL muscle the 25% threshold value was found as the best predictor for actual number of burst estimation. Detailed analyses of the actual and estimated number of bursts had shown that success of threshold estimation may differ among muscle groups. Evaluation of our data had clearly shown that it is important to select proper threshold values for correct EMG signal analyses. Using a single threshold value for different exercise intensities and different muscle groups may cause misleading results.

The correct identification of burst events is crucial in many scenarios, ranging from basic neuroscience to biomedical applications. However, none of the burst detection methods that can be found in the literature have been widely adopted for this task. As an alternative to conventional techniques, a novel neuromorphic approach for real-time burst detection is proposed and tested on acquisitions from in vitro cultures. The system consists of a Neuromorphic Auditory Sensor, which converts the input signal obtained from electrophysiological recordings into spikes and decomposes them into different frequency bands. The output of the sensor is sent to a trained spiking neural network implemented on a SpiNNaker board that discerns between bursting and non-bursting activity. This data-driven approach was compared with 8 different conventional spike-based methods, addressing some of their drawbacks, such as being able to detect both high and low frequency events and working in an online manner. Similar results in terms of number of detected events, mean burst duration and correlation as current state-ofthe-art approaches were obtained with the proposed system, also benefiting * The first two authors (Juan P. Dominguez-Morales and Stefano Buccelli) have contributed equally to this paper.

In this paper we present techniques for detecting and locating transient pipe burst events in water distribution systems. The proposed method uses multiscale wavelet analysis of high rate pressure data recorded to detect transient events. Both wavelet coefficients and Lipschitz exponents provide additional information about the nature of the signal feature detected and can be used for feature classification. A local search method is proposed to estimate accurately the arrival time of the pressure transient associated with a pipe burst event. We also propose a graph-based localization algorithm which uses the arrival times of the pressure transient at different measurement points within the water distribution system to determine the actual location (or source) of the pipe burst. The detection and localization performance of these algorithms is validated through leak-off experiments performed on the WaterWiSe@SG wireless sensor network test bed, deployed on the drinking water distribution system in Singapore. Based on these experiments, we also present a systematic analysis of the sources of localization error. Keywords Multiscale wavelet analysis • transient detection • pipe burst • burst localization 1 Introduction Urban utilities such as drinking water distribution systems (WDSs) are critical infrastructures that increasingly large numbers of residents rely on daily. As populations in cities grow, the demand on these critical infrastructures also grows and the need for real-time monitoring and maintenance becomes vital to ensure efficient, reliable operation and timely response to

Bursts of drinking water pipes not only cause loss of drinking water, but also damage below and above ground infrastructure. Short-term water demand forecasting is a valuable tool in burst detection, as deviations between the forecast and actual water demand may indicate a new burst. Many of burst detection methods struggle with false positives due to non-seasonal water consumption as a result of e.g. environmental, economic or demographic exogenous influences, such as weather, holidays, festivities or pandemics. Finding a robust alternative that reduces the false positive rate of burst detection and does not rely on data from exogenous processes is essential. We present such a burst detection method, based on Bayesian ridge regression and Random Sample Consensus. Our exogenous nowcasting method relies on signals of all nearby flow and pressure sensors in the distribution net with the aim to reduce the false positive rate. The method requires neither data of exogenous processes, nor extensive historical data, but only requires one week of historical data per flow/pressure sensor. The exogenous nowcasting method is compared with a common water demand forecasting method for burst detection and shows sufficiently higher Nash-Sutcliffe model efficiencies of 82.7%-90.6% compared to 57.9%-77.7%, respectively. These efficiency ranges indicate a more accurate water demand prediction, resulting in more precise burst detection.

Automatic and accurate detection of the closure-burst transition events of stops and affricates serves many applications in speech processing. A temporal measure named the plosion index is proposed to detect such events, which are characterized by an abrupt increase in energy. Using the maxima of the pitch-synchronous normalized cross correlation as an additional temporal feature, a rule-based algorithm is designed that aims at selecting only those events associated with the closure-burst transitions of stops and affricates. The performance of the algorithm, characterized by receiver operating characteristic curves and temporal accuracy, is evaluated using the labeled closure-burst transitions of stops and affricates of the entire TIMIT test and training databases. The robustness of the algorithm is studied with respect to global white and babble noise as well as local noise using the TIMIT test set and on telephone quality speech using the NTIMIT test set. For these experiments, th...

PurposeThe purpose of this paper is to identify and discuss the most important research areas on information sharing in supply chains and related risks, taking into account their evolution over time. This paper sheds light on what is happening today and what the trajectories for the future are, with particular respect to the implications for supply chain management.Design/methodology/approachThe dynamic literature review method called Systematic Literature Network Analysis (SLNA) was adopted. It combines the Systematic Literature Review approach and bibliographic network analyses, and it relies on objective measures and algorithms to perform quantitative literature-based detection of emerging topics.FindingsThe focus of the literature seems to be on threats that are internal to the extended supply chain rather than on external attacks, such as viruses, traditionally related to information technology (IT). The main arising risk appears to be the intentional or non-intentional leakage...

It is commonly agreed that one needs to use a threshold value in the detection of muscle activity timing in electromyographic (EMG) signal analysis. However, the algorithm for threshold determination lacks an agreement between the investigators. In this study we aimed to determine a proper threshold value in an incremental cycling exercise for accurate EMG signal analysis. Nine healthy recreationally active male subjects cycled until exhaustion. EMG recordings were performed on four low extremity muscle groups; gastrocnemius lateralis (GL), gastrocnemius medialis (GM), soleus (SOL) and vastus medialis (VM). We have analyzed our data using three different threshold levels: 25%, 35% and 45% of the mean RMS EMG value. We compared the appropriateness of these threshold values using two criteria: (1) significant correlation between the actual and estimated number of bursts and (2) proximity of the regression line of the actual and estimated number of bursts to the line of identity. It had been possible to find a significant correlation between the actual and estimated number of bursts with the 25, 35 and 45% threshold values for the GL muscle. Correlation analyses for the VM muscle had shown that the number of bursts estimated with the 35% threshold value was found to be significantly correlated with the actual number of bursts. For the GM muscle, it had been possible to predict the burst number by using either the 35% or 45% threshold value and for the SOL muscle the 25% threshold value was found as the best predictor for actual number of burst estimation. Detailed analyses of the actual and estimated number of bursts had shown that success of threshold estimation may differ among muscle groups. Evaluation of our data had clearly shown that it is important to select proper threshold values for correct EMG signal analyses. Using a single threshold value for different exercise intensities and different muscle groups may cause misleading results.

In this information age, social media is a powerful online communication tool for people to present their expression such as the real-time events report, personal information including their emotions. Social media text significantly demonstrate information in a current society, also indicate a trend of the dynamic social movement. However, these text characteristics are in a form of informal and unstructured language, i.e. using abbreviation, short text message, slang and argot. It is difficult to classify and extract the key information. There are many techniques proposed to categorize this kind of text information. The social media text classification by using Term Frequency-Inverse Document Frequency (TF-IDF) weighting with Word Article Matrix (WAM) and initial keywords from the well-formed source, i.e. online news article, is one of a productive technique. It generates a set of the sufficient keyword terms to represent text categories. In this paper, we discuss about the proper ...

Social media text can illustrate significant information of our real social situation. It can show the direction of real-time social movement. However, it has its own characteristics such as using short text and informal language, many unstructured information and argot. This kind of text is hard to classify and difficult to analyze to extract the useful information. In this paper, we propose an effective technique to classify the social media text by utilizing the initial keywords from well-formed sources of data, such as online news. Term frequency-inverse document frequency weighting technique (TF-IDF) and Word Article Matrix (WAM) are used as main methods in this research. We use the extracted keywords from the well-formed source as a main factor to do experiment on Twitter messages. We found a set of the social media keywords can represent the essence of social events and can be used to classify the text effectively.

Identifying similarities in microblog posts for event detection poses challenges due to short texts with idiosyncratic spellings, irregular writing styles, abbreviations and synonyms. In order to overcome these challenges, we present an enhancement to the incremental clustering techniques by detecting similar terms in microblog posts in a temporal context. We devise an unsupervised method to measure the similarities online using co-occurrence-based techniques and use them in a vector expansion process. The results of our evaluation performed on a tweet set indicate that the proposed vector expansion method helps identify similarities in tweets despite differences in their content. This facilitates the clustering of tweets and detection of events with higher accuracy without incurring a high execution cost.

The presented thesis investigates the identification of burst locations in water distribution systems (WDS) by analysis of field and simulation experimental data. This required the development of a new hybrid method of burst detection and sizing, and also a burst location identification algorithm. Generally, existing practice relies on a combination of some simple procedure and experience of the involved staff and cannot be easily automated. The practical methods are based on direct manifestation of burst on

The presented thesis investigates the identification of burst locations in water distribution systems (WDS) by analysis of field and simulation experimental data. This required the development of a new hybrid method of burst detection and sizing, and also a burst location identification algorithm. Generally, existing practice relies on a combination of some simple procedure and experience of the involved staff and cannot be easily automated. The practical methods are based on direct manifestation of burst on

The media today bombards us with massive amounts of news about events ranging from the mundane to the memorable. This growing cacophony places an ever greater premium on being able to identify significant stories and to capture their salient features. In this paper, we consider the problem of mining on-line news over a certain period to identify what the major stories were in that time. Major stories are defined as those that were widely reported, persisted for significant duration or had a lasting influence on subsequent stories. Recently, some statistical methods have been proposed to extract important information from large corpora, but most of them do not consider the full richness of language or variations in its use across multiple reporting sources. We propose a method to extract major stories from large news corpora using a combination Latent Dirichlet Allocation and with n-gram analysis.

The proliferation of Internet-enabled smartphones has ushered in an era where events are reported on social media websites such as Twitter and Facebook. However, the short text nature of social media posts, combined with a large volume of noise present in such datasets makes event detection challenging. This problem can be alleviated by using other sources of information, such as news articles, that employ a precise and factual vocabulary, and are more descriptive in nature. In this paper, we propose Spatio-Temporal Event Detection (STED), a probabilistic model to discover events, their associated topics, time of occurrence, and the geospatial distribution from multiple data sources, such as news and Twitter. The joint modeling of news and Twitter enables our model to distinguish events from other noisy topics present in Twitter data. Furthermore, the presence of geocoordinates and timestamps in tweets helps find the spatio-temporal distribution of the events. We evaluate our model ...

Every day 645 million Twitter users generate approximately 58 million tweets. This motivates the question if it is possible to generate a summary of events from this rich set of tweets only. Key challenges in post summarization from microblog posts include circumnavigating spam and conversational posts. In this study, we present a novel technique called lexi-temporal clustering (LTC), which identifies key events. LTC uses k-means clustering and we explore the use of various distance measures for clustering using Euclidean, cosine similarity and Manhattan distance. We collected three original data sets consisting of Twitter microblog posts covering sporting events, consisting of a cricket and two football matches. The match summaries generated by LTC were compared against standard summaries taken from sports sections of various news outlets, which yielded up to 81% precision, 58% recall and 62% F-measure on different data sets. In addition, we also report results of all three variant...

Every day 645 million Twitter users generate approximately 58 million tweets. This motivates the question if it is possible to generate a summary of events from this rich set of tweets only. Key challenges in post summarization from microblog posts include circumnavigating spam and conversational posts. In this study, we present a novel technique called lexi-temporal clustering (LTC), which identifies key events. LTC uses k-means clustering and we explore the use of various distance measures for clustering using Euclidean, cosine similarity and Manhattan distance. We collected three original data sets consisting of Twitter mi-croblog posts covering sporting events, consisting of a cricket and two football matches. The match summaries generated by LTC were compared against standard summaries taken from sports sections of various news outlets, which yielded up to 81% precision, 58% recall and 62% F-measure on different data sets. In addition, we also report results of all three varian...

Nowadays, almost all text corpora, such as blogs, emails and RSS feeds, are a collection of text streams. The traditional vector space model (VSM), or bagof-words representation, cannot capture the temporal aspect of these text streams. So far, only a few bursty features have been proposed to create text representations with temporal modeling for the text streams. We propose bursty feature representations that perform better than VSM on various text mining tasks, such as document retrieval, topic modeling and text categorization. For text clustering, we propose a novel framework to generate bursty distance measure. We evaluated it on UP-GMA, Star and K-Medoids clustering algorithms. The bursty distance measure did not only perform equally well on various text collections, but it was also able to cluster the news articles related to specific events much better than other models.

An algorithm for the detection and location of sudden bursts in water distribution networks combining both continuous monitoring of pressure and hydraulic transient computation is presented. The approach is designed for medium and large bursts that are the result of the sudden rupture of the pipe wall or other physical element in the network and are accompanied by the transient pressure wave that propagates throughout the network. The burst-induced transient wave arrival times and magnitudes measured at two or more points are used to find the location of a burst. The wave arrival times and magnitudes are detected using the modified cumulative sum (CUSUM) change detection test. Results of validation on a real network show the potential of the proposed burst detection and location technique to be used in water distribution systems.

Authors may post the final draft of their work on open, unrestricted Internet sites or deposit it in an institutional repository when the draft contains a link to the bibliographic record of the published version in the ASCE Civil Engineering Database. "Final draft" means the version submitted to ASCE after peer review and prior to copyediting or other ASCE production activities; it does not include the copyedited version, the page proof, or a PDF of the published version

An algorithm for the detection and location of sudden bursts in water distribution networks combining both continuous monitoring of pressure and hydraulic transient computation is presented. The approach is designed for medium and large bursts that are the result of the sudden rupture of the pipe wall or other physical element in the network and are accompanied by the transient pressure wave that propagates throughout the network. The burst-induced transient wave arrival times and magnitudes measured at two or more points are used to find the location of a burst. The wave arrival times and magnitudes are detected using the modified cumulative sum (CUSUM) change detection test. Results of validation on a real network show the potential of the proposed burst detection and location technique to be used in water distribution systems.

Topic detection with large and noisy data collections such as social media must address both scalability and accuracy challenges. KeyGraph is an efficient method that improves on current solutions by considering keyword cooccurrence. We show that KeyGraph has similar accuracy when compared to state-of-the-art approaches on small, well-annotated collections, and it can successfully filter irrelevant documents and identify events in large and noisy social media collections. An extensive evaluation using Amazon's Mechanical Turk demonstrated the increased accuracy and high precision of KeyGraph, as well as superior runtime performance compared to other solutions.

Recently, twitter users are leveraged to detect social and physical events such as festivals and traffic jam at real time. Real time event detection and summarization from Cricket sports is the process of detecting events such as boundary at real time from live Cricket tweet stream as soon as event happens and generating a quick game summary. This is an interesting, yet a complex problem. Because of the need for rapid detection of sports events and for the generation of a concise summary from huge volume of tweets for Cricket enthusiasts. In this paper, a novel framework is proposed for detecting key events from live Cricket tweets and for generating a game summary using the crawled tweets. Feature vectors of live tweets are created using incremental TF-IDF representation and tweet clusters are discovered using Locality Sensitive Hashing (LSH) where the post rate of each cluster determines the key event. A key event is recognized from that cluster using our domain specific event lexicon. Then, important moments from the crawled tweets are computed by identifying the spikes in the tweets volume. Topk tweets from each moment are selected by ranking tweets on top-k words. Representative tweets from top-k tweets are identified using Jaccard similarity. The evaluation on 2017 IPL T20 Cricket live tweets using ROC measure shows that the proposed incremental TF-IDF based LSH approach detects key events with nearly 95% true positive rate and around 5% false positive rate. The proposed game summarization algorithm generates summaries which are readable and competitive to human tailored summaries.

Nowadays, trends detection is an important task on social media to determine trends that are being discussed the most on a social platform. One of the main challenges of this task is the processing of unstructured textual data which has different representations. Therefore, multi view text clustering presents a useful solution for trends detection by integrating various representations called 'views' to provide a complementary description of the same content. In this context, we propose a new ensemble method for multi-view text clustering that exploits different representations of text in order to produce more accurate and high quality clustering. Extensive experiments on real-world text datasets were conducted to demonstrate its superiority by comparing with the existing methods. An application of the proposed method in trends detection from twitter is also illustrated.

is paper presents an approach to automatically detecting breaking news events from social media streams, using event detection to collect in near real time relevant video documents from social networks regarding that breaking news. A visual analytics dashboard provides access to the results of the content processing pipeline, providing a rich interactive interface to explore emerging stories and select video material around those stories for veri cation.

This paper discusses the work done by a team at the University of Florida for the TREC 2011 Microblog Track. To build a real-time microblog search engine we rely on topic modeling for our search. To facilicate our algorithms we bundle similar tweets together in what we call supertweet generation. We perform online inference and offline inference depending on the time frame of the topical query. In this paper we discuss our techniques, challenges, future work, but not the evaluation of our results. We offer this notebook entry to invite future discussion and collaboration.

This paper discusses the work done by a team at the University of Florida for the TREC 2011 Microblog Track. To build a real-time microblog search engine we rely on topic modeling for our search. To facilicate our algorithms we bundle similar tweets together in what we call supertweet generation. We perform online inference and offline inference depending on the time frame of the topical query. In this paper we discuss our techniques, challenges, future work, but not the evaluation of our results. We offer this notebook entry ...

Image tweets are becoming a prevalent form of social media, but little is known about their content-textual and visual-and the relationship between the two mediums. Our analysis of image tweets shows that while visual elements certainly play a large role in image-text relationships, other factors such as emotional elements, also factor into the relationship. We develop Visual-Emotional LDA (VELDA), a novel topic model to capture the image-text correlation from multiple perspectives (namely, visual and emotional). Experiments on real-world image tweets in both English and Chinese and other user generated content, show that VELDA significantly outperforms existing methods on cross-modality image retrieval. Even in other domains where emotion does not factor in image choice directly, our VELDA model demonstrates good generalization ability, achieving higher fidelity modeling of such multimedia documents.

In this paper we present techniques for detecting and locating transient pipe burst events in water distribution systems. The proposed method uses multiscale wavelet analysis of high rate pressure data recorded to detect transient events. Both wavelet coefficients and Lipschitz exponents provide additional information about the nature of the signal feature detected and can be used for feature classification. A local search method is proposed to estimate accurately the arrival time of the pressure transient associated with a pipe burst event. We also propose a graph-based localization algorithm which uses the arrival times of the pressure transient at different measurement points within the water distribution system to determine the actual location (or source) of the pipe burst. The detection and localization performance of these algorithms is validated through leak-off experiments performed on the WaterWiSe@SG wireless sensor network test bed, deployed on the drinking water distribution system in Singapore. Based on these experiments, we also present a systematic analysis of the sources of localization error.

Web mining -is the application of data mining techniques to discover patterns from the Web. Topic tracking is one of the technologies that has been developed and can be used in the text mining process. The main purpose of topic tracking is to identify and follow events presented in multiple news sources, including newswires, radio and TV broadcasts. In this paper, a survey of topic tracking techniques is presented.

Textual topic detection methods that work by clustering terms according to their cooccurrence patterns are called feature-pivot methods. Typically, the similarity measure that is used for such clustering methods takes into account the cooccurrence patterns of only pairs of items. In this work, we argue that examining the simultaneous cooccur-rence patterns of a larger number of terms, is a better option when the corpus contains a set of closely related fine-grained topics. To this end, we treat the topic detection problem as a Frequent Pattern Mining problem and propose a novel algorithm for " soft " Frequent Pattern Mining. We test the proposed approach using three annotated datasets collected from Twitter and compare it to a set of algorithms that includes a graph-based feature-pivot approach that takes into account only cooccurrence patterns, a standard Frequent Pattern Mining algorithm and Latent Dirichlet Allocation. The results indicate that SFPM is performing better than the other tested methods and show a clear improvement over the standard FPM approach.

Twitter is the most popular micro blogging web site. More than millions of tweets are posted along twitter every day. Tweets contains huge amount of noisy and redundant data. It is very important to summarize the huge amount of tweets by reducing the size of tweets and removing the noise, for improving the result accuracy. The operations over flood of tweets are not an easy task. There are so many tweets are unrelated, also arrival rate of tweets is fast. To handle these problems, there is a need of efficient and strong summarization algorithm. This algorithm should be flexible with random time duration. For topic evolution system should detect sub-topic and keeps track for any changes occur with the time. To achieve all these goals, proposed system performs three types of operations on tweets, named as clustering of tweets, summarization and topic evaluation over tweeter data. Framework has component is data duplication checking using SHA1 hashing strategy. Framework used clustering procedure it uses EM clustering and compare the EM clustering algorithm with K-means clustering algorithm. After this, tweets are summarized with greedy algorithm, which is more accuracy as compared to traditional summarization algorithm. Finally, the topic is detected for generated summary. Experimental results proves that the proposed system summarize the tweets more accurately and efficiently.

Tweet are being created short text message and shared for both users and data analysts. Twitter which receives over 400 million tweets per day has emerged as an invaluable source of news, blogs, opinions and more. Our proposed work consists three components tweet stream clustering to cluster tweet using k-means cluster algorithm and second tweet cluster vector technique to generate rank summarization using greedy algorithm, therefore requires functionality which significantly differ from traditional summarization. in general, tweet summarization and third to detect and monitors the summary-based and volume based variation to produce timeline automatically from tweet stream. Implementing continuous tweet stream reducing a text document is however not an simple task, since a huge number of tweets are worthless, unrelated and raucous in nature, due to the social nature of tweeting. Further, tweets are strongly correlated with their posted instance and up-to-the-minute tweets tend to arrive at a very fast rate. Efficiency—tweet streams are always very big in level, hence the summarization algorithm should be greatly capable; Flexibility—it should provide tweet summaries of random moment durations. (3) Topic evolution—it should routinely detect sub-topic changes and the moments that they happen.

www.erpublication.org  Abstract-In recent days usage of social networking sites like twitter, face book and Instagram has become so popular that they allows us to become informed on various aspects of companies, news, people, their views and emotions. Tweets on twitter arrive at a very high rate of 6000 tweets per second and are created in unorganized and informal way with lots of noise and redundancy. The searching process for concerned 140 characters of short message (tweet) out of the huge data collection becomes a hectic task to both users and data analysts. The best solution which we can frame for this problem is to apply continuous summarization technique on it with TCV Rank summarization algorithm after initially clustering the tweet streams using k-means clustering. Summarization along with automatic timeline generation with the topic evolution detection method algorithm minimizes the efforts. This technique when implemented will make the work of data analysts and also end user easier specially in searching important tweets. This paper mainly focuses on efficiency, flexibility and topic evolution factors.

Twitter is one of the famous microblogging services, with hundreds of millions of tweets being posted every day on a wide variety of topics. Tweets are the short text messages which are limited upto 140 characters in length. They are generated and given out at an unusual rate. Tweets are descriptive in nature. Tweets contains too many noisy data and redundancies. In this , we make an attempt to introduce a distinctive new framework for continuous summarization called Summblr to deal with the problem. Existing methods that summarize the documents highlight on smallscale data sets that are static. Summblr, in contrast,is developed to deal with large-scale tweet data streams which arrive at a faster rate dynamically. Our constructed framework includes three major components. As a first step, a clustering algorithm for tweet data stream is proposed which is online to cluster togetherly the tweets and maintain it in one of the data structure called as TCV that is tweet cluster vector . Secondarily, a novel technique called TCV-Rank summarization for generating summaries both online and historical of any time durations is proposed.

In this paper we provide context to Topic Modelling as an Information Warfare technique. Topic modelling is a technique that discovers latent topics in unstructured and unlabelled collection of documents. The topic structure can be searched for interesting and relevant topics. The objectives of this paper is to describe topic modelling, put it in context as a useful IW technique and illustrate its use with two examples. We discuss several applications of topic modelling in the safety and security domain and list several topic model variations that are of special interest to the IW domain.

First Story Detection is hard because the most accurate systems become progressively slower with each document processed. We present a novel approach to FSD, which operates in constant time/space and scales to very high volume streams. We show that when computing novelty over a large dataset of tweets, our method performs 192 times faster than a state-of-the-art baseline without sacrificing accuracy. Our method is capable of performing FSD on the full Twitter stream on a single core of modest hardware.

Over the last 40 years, automatic solutions to analyze text documents collection have been one of the most attractive challenges in the field of information retrieval. More recently, the focus has moved towards dynamic, distributed environments, where documents are continuously created by the users of a virtual community, i.e., the social network. In the case of Twitter, such documents, called tweets, are usually related to events which involve many people in different parts of the world. In this work we present a system for real-time Twitter data analysis which allows to follow a generic event from the user's point of view. The topic detection algorithm we propose is an improved version of the Soft Frequent Pattern Mining algorithm, designed to deal with dynamic environments. In particular, in order to obtain prompt results, the whole Twitter stream is split in dynamic windows whose size depends both on the volume of tweets and time. Moreover, the set of terms we use to query Twitter is progressively refined to include new relevant keywords which point out the emergence of new subtopics or new trends in the main topic. Tests have been performed to evaluate the performance of the framework and experimental results show the effectiveness of our solution.

Social media text can illustrate significant information of our real social situation. It can show the direction of real-time social movement. However, it has its own characteristics such as using short text and informal language, many unstructured information and argot. This kind of text is hard to classify and difficult to analyze to extract the useful information. In this paper, we propose an effective technique to classify the social media text by utilizing the initial keywords from well-formed sources of data, such as online news. Term frequency–inverse document frequency weighting technique (TF-IDF) and Word Article Matrix (WAM) are used as main methods in this research. We use the extracted keywords from the well-formed source as a main factor to do experiment on Twitter messages. We found a set of the social media keywords can represent the essence of social events and can be used to classify the text effectively.

An algorithm for the detection and location of sudden bursts in water distribution networks combining both continuous monitoring of pressure and hydraulic transient computation is presented. The approach is designed for medium and large bursts that are the result of the sudden rupture of the pipe wall or other physical element in the network and are accompanied by the transient pressure wave that propagates throughout the network. The burst-induced transient wave arrival times and magnitudes measured at two or more points are used to find the location of a burst. The wave arrival times and magnitudes are detected using the modified cumulative sum (CUSUM) change detection test. Results of validation on a real network show the potential of the proposed burst detection and location technique to be used in water distribution systems.

An algorithm for the burst detection and location in water distribution networks based on the continuous monitoring of the flow rate at the entry point of the network and the pressure at a number of points within the network is presented. The approach is designed for medium to large bursts with opening times in the order of a few minutes and is suitable for networks of relatively small size, such as district metered areas (DMAs). The burst-induced increase in the inlet flow rate is detected using the modified cumulative sum (CUSUM) change detection test. Based on parameters obtained from the CUSUM test, the burst is simulated at a number of burst candidate locations. The calculated changes in pressure at the pressure monitoring points are then compared to the measured values and the location resulting in the best fit is selected as the burst location. The EPANET steady-state hydraulic solver is utilised to simulate the flows and pressures in the network. A sensitivity-based sampling design procedure is introduced to find the optimal positions for pressure monitoring points. The proposed algorithm is tested on a case study example network and shows potential for burst detection and location in real water distribution systems.

This paper investigates the potential of using controlled hydraulic transients for non-intrusive assessment of the internal condition of water transmission pipelines. Deterioration of pipelines is a natural process. An effective tool for evaluating the state of pipeline interiors could thus be extremely useful for planning rehabilitation or identifying critical points in a system. A non-intrusive technique for evaluating the state of the internal surface of a pipeline is introduced. The total length of the pipeline was divided into a number of segments and the condition of each segment assessed based on the transient response corresponding to that interval. In addition to quantification of the deterioration level of different sections of the pipeline, critical points were identified where closer investigation was suggested. The proposed technique, although in an early stage of development, was tested on a real pipeline.

Textual topic detection methods that work by clustering terms according to their cooccurrence patterns are called feature-pivot methods. Typically, the similarity measure that is used for such clustering methods takes into account the cooccurrence patterns of only pairs of items. In this work, we argue that examining the simultaneous cooccurrence patterns of a larger number of terms, is a better option when the corpus contains a set of closely related fine-grained topics. To this end, we treat the topic detection problem as a Frequent Pattern Mining problem and propose a novel algorithm for "soft" Frequent Pattern Mining. We test the proposed approach using three annotated datasets collected from Twitter and compare it to a set of algorithms that includes a graph-based feature-pivot approach that takes into account only cooccurrence patterns, a standard Frequent Pattern Mining algorithm and Latent Dirichlet Allocation.

An algorithm for the detection and location of sudden bursts in water distribution networks combining both continuous monitoring of pressure and hydraulic transient computation is presented. The approach is designed for medium and large bursts that are the result of the sudden rupture of the pipe wall or other physical element in the network and are accompanied by the transient

ABSTRACT
The web has become the fastest growing and the most up to date source of information. Web mining is extracting knowledge from data available on the web by applying data mining techniques. News articles are being generated by various news agencies every day on the web. Articles available online is more easily accessible and most often cited by the users. Event detection involves monitoring the continuous stream of news stories to identify events which went unnoticed previously in the news stream and explore new events. In this paper, various techniques for detecting events in the broadcast Newswire stories are analyzed.
Keywords:- Web mining, Online new event detection, News streams, Topic detection and tracking (TDT).

A new approach to gas leakage detection in high pressure distribution networks is proposed, where two leakage detectors are modelled as a Linear Parameter Varying (LPV) system whose scheduling signals are, respectively, intake and offtake pressures. Running the two detectors simultaneously allows for leakage location. First, the pipeline is identified from operational data, supplied by REN-Gasodutos and using an LPV systems identification algorithm proposed in . Each leakage detector uses two Kalman filters where the fault is viewed as an augmented state. The first filter estimates the flow using a calculated scheduling signal, assuming that there is no leakage. Therefore it works as a reference. The second one uses a measured scheduling signal and the augmented state is compared with the reference value. Whenever there is a significant difference, a leakage is detected. The effectiveness of this method is illustrated with an example where a mixture of real and simulated data is used.

In this work a high pressure distribution pipeline is modelled as a linear parameter varying (LPV) system. The pipeline is afterwards identified from operational data, supplied by RENGasodutos. On the basis of the LPV model, a leakage detector and locator was build and it showed good performance for 10% leakages.