Document Clustering - Concepts, Metrics and Algorithms

Availability of large full-text document collections in electronic form has created a need for tools and techniques that assist users in organizing these collections. Document clustering is one of the popular methods used for this purpose. The Self-organizing map (SOM), an unsupervised algorithm for clustering and topographic mapping, has shown promising results in this task. Most of the existing SOM techniques rely on a “bag of words” document representation. Each word in the document is considered as a separate feature, ignoring the word order. In this chapter we investigate the use of phrases rather than words as document features for document clustering. We present a phrase grammar extraction technique, and use the extracted phrases as features in two different document clustering algorithms, self-organizing map (SOM) and hierarchical self-organizing map (HSOM). We present results of clustering documents from the REUTERS corpus and show an improvement in the clustering performance evaluated using the entropy and F-measure.

Clustering is a technique that group data together based on their similarity and apart based on their dissimilarity. When this technique is applied to documents and the terms within these documents retrieval of similar documents become easy and efficient. Document clustering is being researched and utilized for many years but is yet far from being optimal. To study and analyze different document clustering algorithm, a theoretical literature review and analysis was performed and the results are presented in this paper. This paper comprises of theoretical review of papers. 95 papers were identified and out of these 30 were selected. Various techniques or algorithms and modifications to previous algorithms proposed for document clustering by various researchers are compiled and presented with the intent that it will aid the researchers in finding out the current and future scope of research in information retrieval systems and document clustering technologies.

Clustering means that identical items are held together. Document clustering is an application of clustering, which refers to having related text records together. Document clustering plays a crucial role in the creation of search engines, where a document category is expected to be listed in the minimum answer period as a result of the query. Document clustering is critical in terms of the final objective of tracking membership, rundown, collection of subjects and data retrieval in a competent manner. First the designation is related to the upgrading of data recovery procedures. Recently, clustering techniques have been related in the regions, which involve browsing the gathered knowledge or ordering the findings of the web indices to address the query posed by the clients. This paper elaborates on the idea of document cum text clustering. This paper would include a survey of recent work in the area of text clustering. This paper would also include a critical analysis of current text clustering techniques.

This paper presents a novel document clustering algorithm that represents documents as a time series of words. Document clustering is very important due to the fact that it permits us to group them based on some certain criteria, especially nowadays when a large number of articles are available. The timed series representation of the document instead of the vector model permits us to consider a new algorithm for the computation of the distance between documents: dynamic time warping. This novel representation together with the dynamic time warping algorithm represents the foundation for computing the similarity and the clustering of the documents. The clustering algorithm used is hierarchical clustering. This novel clustering method of texts is applied on named entities and on the parts of speech of the words that compose the documents. As test data we are using the Reuters corpus of newspaper articles.

Document clustering techniques mostly rely on single term analysis of the document data set, such as the vector space model. To achieve more accurate document clustering, more informative features including phrases and their weights are particularly important in such scenarios. Document clustering is particularly useful in many applications such as automatic categorization of documents, grouping search engine results, building a taxonomy of documents, and others. This article presents two key parts of successful document clustering. The first part is a novel phrase-based document index model, the document index graph, which allows for incremental construction of a phrase-based index of the document set with an emphasis on efficiency, rather than relying on single-term indexes only. It provides efficient phrase matching that is used to judge the similarity between documents. The model is flexible in that it could revert to a compact representation of the vector space model if we choose not to index phrases. The second part is an incremental document clustering algorithm based on maximizing the tightness of clusters by carefully watching the pair-wise document similarity distribution inside clusters. The combination of these two components creates an underlying model for robust and accurate document similarity calculation that leads to much improved results in Web document clustering over traditional methods.

Document clustering is a very useful application in recent days especially with the advent of the World Wide Web. Most of the existing document clustering algorithms either produce clusters of poor quality or are highly computationally expensive. In this paper we propose a document-clustering algorithm, KMART, that uses an unsupervised fuzzy adaptive resonance theory (fuzzy-ART) neural network. A modified version of the fuzzy ART is used to enable a document to be in multiple clusters. The number of clusters is determined dynamically. Some experiments are reported to compare the efficiency and execution time of our algorithm with other document-clustering algorithm like fuzzy c-means. The results show that KMART is both effective and efficient.

Document clustering organizes documents into groups such that each group contains documents with similar content. The majority of document clustering algorithms require a vector representation for each document. Each vector has well over 10,000 elements. Consequently, the memory required during clustering can be extremely high when clustering hundreds of thousands of documents. This paper introduces document vector compression, which is based on the discrete cosine transform (DCT). Document vector compression reduces the run-time memory requirements by as much as 60%. Document vector compression does not degrade the final cluster quality (total F-measure) as does other document vector reduction techniques

Document clustering, which involves concepts from the fields of information retrieval, automatic topic extraction, natural language processing, and machine learning, is one of the most popular research areas in data mining. Due to the large amount of information in electronic form, fast and high-quality cluster analysis plays an important role in helping users to effectively navigate, summarize and organise this information for useful data. There are a number of techniques in the literature, which efficiently provide solutions for document clustering. However, during the last decade, researchers started to use metaheuristic algorithms for the document clustering problem because of the limitations of the existing traditional clustering algorithms. In this chapter, the authors will give a brief review of various research papers that present the area of document or text clustering approaches with different metaheuristic algorithms.

With the rapid development of the Internet, huge volumes of text data are also increasing. These rapidly growing data generate challenges for users such as access, organizing, and analyzing the required information expeditiously. Document clustering techniques mostly rely on the statistical analysis of a term. It can be hard to identify in situations when multiple terms have the same frequency value, but one term is more important in terms of meaning than the other. Also, the process to discover more relevant information regarding a user query on the Web is uncontrollable. The proposed system tries to implement a concept-based document clustering model that clusters the Web documents based on the semantics or theme of the text data. The semantic analysis is done with the help of the semantic role labeler (SRL) to find the terms that contribute more to the meaning of the sentence. This system is called a concept-based analysis mechanism (CBAM). This system can provide accurate and robust document similarity calculation as well as concept identification that lead to improved results in Web document clustering.

This paper introduces a prototype web search results clustering engine that enhances search results by performing fuzzy clustering on web documents returned by conventional search engines, as well as ranking the results and labeling the resulting clusters. This is done using a fuzzy transduction-based clustering algorithm (FTCA), which employs a transduction-based relevance model (TRM) to generate document relevance values. These relevance values are used to cluster similar documents, rank them, and facilitate a term frequency based label generator. The membership degrees of documents to fuzzy clusters also facilitates effective detection and removal of overly similar clusters. FTCA is compared against two other established web document clustering algorithms: Suffix Tree Clustering (STC) and Lingo, which are provided by the free open source Carrot <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> Document Clustering Workbench. To measure cluster quality, an extended version of the classic precision measurement is used to take into account relevance and fuzzy clustering, along with recall and F1 score. Results from testing on five different datasets show a considerable clustering quality and performance advantage over STC and Lingo in most cases.

Document clustering plays a vital role in text mining fields such as information retrieval, sentiment analysis, and text organizing. Document clustering aims to automatically divide a collection of documents based on some aspects of similarity into groups that are meaningful, useful or both. This paper aims to improve the clustering task for the Arabic documents. Recent studies show that partitioning clustering algorithms are more suitable for clustering process. However, k-means is the most common algorithm that is being used for clustering process because of its simplicity and speed. It can only generate an arbitrary solution because the results depend on the initial centers for the desired clusters “the seeds”. In this paper, a new modified k-means algorithm called PSO K-means, supported by Particle Swarm Optimization (PSO) is applied to enhance the Arabic document clustering process. Then, an intensive comparative study between the proposed model and the standard k-means algorithm is applied. Also, the stemming algorithms those are being used in Arabic language processing were assessed. Through the experiments, an evaluation for the new algorithm is done with three different Arabic data sets. The results demonstrate that the proposed model can produce more accurate results compared to the standard k-means algorithm for Arabic language documents. On the other hand, Arabic light stemmer is more suitable for the stemming step.

Clustering is a division of data into groups of similar objects. Each group, called cluster, consists of objects that are similar between themselves and dissimilar to objects of other groups. The clustering algorithm attempts to find natural groups of components, based on some similarity. Traditional clustering algorithms will search only a small sub-set of all possible clustering and consequently, there is no guarantee that the solution found will be optimal. This paper presents the document clustering based on Genetic algorithm with Simultaneous mutation operator and Ranked mutation rate. The mutation operation is significant to the success of genetic algorithms since it expands the search directions and avoids convergence to local optima. In each stage of the genetic process in a problem, may involve aptly different mutation operators for best results. In simultaneous mutation the genetic algorithm concurrently uses several mutation operators in producing the next generation. The mutation ratio of each operator changes according to assessment from the respective offspring it produces. In ranked scheme, it adapts the mutation rate on the chromosome based on the fitness rank of the earlier population. Experiments results are examined with document corpus. It demonstrates that the proposed algorithm statistically outperforms the Simple GA and K-Means.

As an unsupervised machine learning technology, document clustering has been widely used in many fields, such as Information Retrieval (IR) and Text Categorization (TC). But, because of the bag of words used in document clustering as document index, the feature space of corpus must be high dimension space. This problem makes a negative effect to the efficiency and precision of text clustering. Based on category resolve power, a new feature selection function is constructed. Through integration between this function and document clustering algorithm, a high-powered text clustering algorithm is presented. Experiments on a universal corpus show that it has a good performance.

Text Document Clustering is one of the fastest growing research areas because of availability of huge amount of information in an electronic form. There are several number of techniques launched for clustering documents in such a way that documents within a cluster have high intra-similarity and low inter-similarity to other clusters. Many document clustering algorithms provide localized search in effectively navigating, summarizing, and organizing information. A global optimal solution can be obtained by applying high-speed and high-quality optimization algorithms. The optimization technique performs a globalized search in the entire solution space. In this paper, a brief survey on optimization approaches to text document clustering is turned out.

Document clustering is an aggregation of related documents to a cluster based on the similarity evaluation task between documents and the representatives of clusters. Terms and their discriminating features of terms are the clue to the clustering and the discriminating features are based on the term and document frequencies. Feature selection method on the basis of frequency statistics has a limitation to the enhancement of the clustering algorithm because it does not consider the contents of the cluster objects. In this paper, we adopt a content-based analytic approach to refine the similarity computation and propose a keyword-based clustering algorithm. Experimental results show that content-based keyword weighting outperforms frequency-based weighting method.