Text clustering with important words using normalization

Important words, which usually exist in part of Title, Subject and Keywords, can briefly reflect the main topic of a document. In recent years, it is a common practice to exploit the semantic topic of documents and utilize important words to achieve document clustering, especially for short texts such as news articles. This paper proposes a novel method to extract important words from Subject and Keywords of articles, and then partition documents only with those important words. Considering the fact that frequencies of important words are usually low and the scale matrix dataset for important words is small, a normalization method is then proposed to normalize the scale dataset so that more accurate results can be achieved by sufficiently exploiting the limited information. The experiments validate the effectiveness of our method.

Mining streams of short text for analysis of world-wide event evolutions

Streams of short text, such as news titles, enable us to effectively and efficiently learn the real world events that occur anywhere and anytime. Short text messages that are companied by timestamps and generally brief events using only a few words differ from other longer text documents, such as web pages, news stories, blogs, technical papers and books. For example, few words repeat in the same news titles, thus frequency of the term (i.e., TF) is not as important in short text corpus as in longer text corpus. Therefore, analysis of short text faces new challenges. Also, detecting and tracking events through short text analysis need to reliably identify events from constant topic clusters; however, existing methods, such as Latent Dirichlet Allocation (LDA), generates different topic results for a corpus at different executions. In this paper, we provide a Finding Topic Clusters using Co-occurring Terms (FTCCT) algorithm to automatically generate topics from a short text corpus, and develop an Event Evolution Mining (EEM) algorithm to discover hot events and their evolutions (i.e., the popularity degrees of events changing over time). In FTCCT, a term (i.e., a single word or a multiple-words phrase) belongs to only one topic in a corpus. Experiments on news titles of 157 countries within 4 months (from July to October, 2013) demonstrate that our FTCCT-based method (combining FTCCT and EEM) achieves far higher quality of the event's content and description words than LDA-based method (combining LDA and EEM) for analysis of streams of short text. Our method also visualizes the evolutions of the hot events. The discovered world-wide event evolutions have explored some interesting correlations of the world-wide events; for example, successive extreme weather phenomenon occur in different locations - typhoon in Hong Kong and Philippines followed hurricane and storm flood in Mexico in September 2013. © 2014 Springer Science+Business Media New York.

Fast growing self organizing map for text clustering

This paper presents an integration of a novel document vector representation technique and a novel Growing Self Organizing Process. In this new approach, documents are represented as a low dimensional vector, which is composed of the indices and weights derived from the keywords of the document.

An index based similarity calculation method is employed on this low dimensional feature space and the growing self organizing process is modified to comply with the new feature representation model.

The initial experiments show that this novel integration outperforms the state-of-the-art Self Organizing Map based techniques of text clustering in terms of its efficiency while preserving the same accuracy level.

A sequence based dynamic SOM model for text clustering

Text clustering can be considered as a four step process consisting of feature extraction, text representation, document clustering and cluster interpretation. Most text clustering models consider text as an unordered collection of words. However the semantics of text would be better captured if word sequences are taken into account.

In this paper we propose a sequence based text clustering model where four novel sequence based components are introduced in each of the four steps in the text clustering process.

Experiments conducted on the Reuters dataset and Sydney Morning Herald (SMH) news archives demonstrate the advantage of the proposed sequence based model, in terms of capturing context with semantics, accuracy and speed, compared to clustering of documents based on single words and n-gram based models.

Dirichlet process mixture models with pairwise constraints for data clustering

The Dirichlet process mixture (DPM) model, a typical Bayesian nonparametric model, can infer the number of clusters automatically, and thus performing priority in data clustering. This paper investigates the influence of pairwise constraints in the DPM model. The pairwise constraint, known as two types: must-link (ML) and cannot-link (CL) constraints, indicates the relationship between two data points. We have proposed two relevant models which incorporate pairwise constraints: the constrained DPM (C-DPM) and the constrained DPM with selected constraints (SC-DPM). In C-DPM, the concept of chunklet is introduced. ML constraints are compiled into chunklets and CL constraints exist between chunklets. We derive the Gibbs sampling of the C-DPM based on chunklets. We further propose a principled approach to select the most useful constraints, which will be incorporated into the SC-DPM. We evaluate the proposed models based on three real datasets: 20 Newsgroups dataset, NUS-WIDE image dataset and Facebook comments datasets we collected by ourselves. Our SC-DPM performs priority in data clustering. In addition, our SC-DPM can be potentially used for short-text clustering.

Managing email overload with an automatic nonparametric clustering approach

Email overload is a recent problem that there is increasingly difficulty people have faced to process the large number of emails received daily. Currently this problem becomes more and more serious and it has already affected the normal usage of email as a knowledge management tool. It has been recognized that categorizing emails into meaningful groups can greatly save cognitive load to process emails and thus this is an effective way to manage email overload problem. However, most current approaches still require significant human input when categorizing emails. In this paper we develop an automatic email clustering system, underpinned by a new nonparametric text clustering algorithm. This system does not require any predefined input parameters and can automatically generate meaningful email clusters. Experiments show our new algorithm outperforms existing text clustering algorithms with higher efficiency in terms of computational time and clustering quality measured by different gauges.

Sparse subspace clustering via group sparse coding

We propose in this paper a novel sparse subspace clustering method that regularizes sparse subspace representation by exploiting the structural sharing between tasks and data points via group sparse coding. We derive simple, provably convergent, and computationally efficient algorithms for solving the proposed group formulations. We demonstrate the advantage of the framework on three challenging benchmark datasets ranging from medical record data to image and text clustering and show that they consistently outperforms rival methods.

Sensor-cloud computing : novel applications and research problems

Recent developments in sensor networks and cloud computing saw the emergence of a new platform called sensor-clouds. While the proposition of such a platform is to virtualise the management of physical sensor devices, we are seeing novel applications been created based on a new class of social sensors. Social sensors are effectively a human-device combination that sends torrent of data as a result of social interactions and social events. The data generated appear in different formats such as photographs, videos and short text messages. Unlike other sensor devices, social sensors operate on the control of individuals via their mobile devices such as a phone or a laptop. And unlike other sensors that generate data at a constant rate or format, social sensors generate data that are spurious and varied, often in response to events as individual as a dinner outing, or a news announcement of interests to the public. This collective presence of social data creates opportunities for novel applications never experienced before. This paper discusses such applications as a result of utilising social sensors within a sensor-cloud environment. Consequently, the associated research problems are also presented.

Finding short patterns to classify text documents

Many classification methods have been proposed to find patterns in text documents. However, according to Occam's razor principle, "the explanation of any phenomenon should make as few assumptions as possible", short patterns usually have more explainable and meaningful for classifying text documents. In this paper, we propose a depth-first pattern generation algorithm, which can find out short patterns from text document more effectively, comparing with breadth-first algorithm

Text document clustering with metric learning

One reason for semi-supervised clustering fail to deliver satisfactory performance in document clustering is that the transformed optimization problem could have many candidate solutions, but existing methods provide no mechanism to select a suitable one from all those candidates. This paper alleviates this problem by posing the same task as a soft-constrained optimization problem, and introduces the salient degree measure as an information guide to control the searching of an optimal solution. Experimental results show the effectiveness of the proposed method in the improvement of the performance, especially when the amount of priori domain knowledge is limited.

Enhancing an evolving tree-based text document visualization model with fuzzy c-Means clustering

An improved evolving model, i.e., Evolving Tree (ETree) with Fuzzy c-Means (FCM), is proposed for undertaking text document visualization problems in this study. ETree forms a hierarchical tree structure in which nodes (i.e., trunks) are allowed to grow and split into child nodes (i.e., leaves), and each node represents a cluster of documents. However, ETree adopts a relatively simple approach to split its nodes. Thus, FCM is adopted as an alternative to perform node splitting in ETree. An experimental study using articles from a flagship conference of Universiti Malaysia Sarawak (UNIMAS), i.e., Engineering Conference (ENCON), is conducted. The experimental results are analyzed and discussed, and the outcome shows that the proposed ETree-FCM model is effective for undertaking text document clustering and visualization problems.

Multi-label classification with clustering for image and text categorization

This paper explores effective multi-label classification methods for multi-semantic image and text categorization. We perform an experimental study of clustering based multi-label classification (CBMLC) for the target problem. Experimental evaluation is conducted for identifying the impact of different clustering algorithms and base classifiers on the predictive performance and efficiency of CBMLC. In the experimental setting, three widely used clustering algorithms and six popular multi-label classification algorithms are used and evaluated on multi-label image and text datasets. A multi-label classification evaluation metrics, micro F1-measure, is used for presenting predictive performances of the classifications. Experimental evaluation results reveal that clustering based multi-label learning algorithms are more effective compared to their non-clustering counterparts.

The rapid analysis of heroin drug seizures using micellar electrokinetic chromatography with short-end injection

A simple and rapid method for the analysis of heroin seizures by micellar electrokinetic chromatography with short-end injection is described. Separations were performed using an uncoated fused silica capillary, 50 cm×50 mm I.D.×360 mm O.D. with an effective separation length of 8 cm. The system was run at 25°C with an applied negative voltage of –25 kilovolts. Injection of each sample was for 2 s at –50 mbar. UV detection was employed with the wavelength set at 210 nm. The background electrolyte consisted of 85:15 (water:acetonitrile, v/v) containing final concentrations of 25 mM SDS and 15 mM sodium borate, pH 9.5. Samples and standards were prepared in 0.1% v/v acetic acid and diluted in the run buffer containing 1 mg/ml of N,N-dimethyl-5-methoxytryptamine as an internal standard. Under these conditions a text mixture containing caffeine, paracetamol, morphine, codeine, heroin, and acetylcodeine was resolved within 1.5 min. The method was used to determine the concentration of heroin in heroin seizure samples, and the results were in good agreement with those obtained by a validated gas chromatographic method.

Enhancing the effectiveness of clustering with spectra analysis

For many clustering algorithms, such as K-Means, EM, and CLOPE, there is usually a requirement to set some parameters. Often, these parameters directly or indirectly control the number of clusters, that is, k, to return. In the presence of different data characteristics and analysis contexts, it is often difficult for the user to estimate the number of clusters in the data set. This is especially true in text collections such as Web documents, images, or biological data. In an effort to improve the effectiveness of clustering, we seek the answer to a fundamental question: How can we effectively estimate the number of clusters in a given data set? We propose an efficient method based on spectra analysis of eigenvalues (not eigenvectors) of the data set as the solution to the above. We first present the relationship between a data set and its underlying spectra with theoretical and experimental results. We then show how our method is capable of suggesting a range of k that is well suited to different analysis contexts. Finally, we conclude with further  empirical results to show how the answer to this fundamental question enhances the clustering process for large text collections.