Recent Developments in Document Clustering

2019

2015

Data mining , knowledge discovery is the process of analyzing data from different perspectives and summarizing it into useful information information that can be used to increase revenue, cuts costs, or both. Data mining software is one of a number of analytical tools for analyzing data. It allows users to analyze data from many different dimensions or angles, categorize it, and summarize the relationships identified. Technically, data mining is the process of finding correlations or patterns among dozens of fields in large relational databases. The goal of clustering is to determine the intrinsic grouping in a set of unlabeled data. But how to decide what constitutes a good clustering? It can be shown that there is no absolute “best” criterion which would be independent of the final aim of the clustering. Consequently, it is the user which must supply this criterion, in such a way that the result of the clustering will suit

The amount of digital data utilized in daily life has increased owing to the high dependence on such data. Most data can be stored in textual documents. With the rapid increase in the number of textual documents, users face problems in obtaining useful information. Thus, a method by which to manage data is required to give users an idea about content. In addition, techniques to increase the ratio of precision in information retrieval results are also needed. Therefore, the textual document clustering area is developed to represent the data in meaningful clusters. The two main factors encountered in the process of textual document clustering are efficiency and goodness or quality of data clusters. Efforts have been exerted to deal with these factors. These attempts can be categorized into either traditional or modern approaches. However, these attempts also face numerous issues. In this paper, we present the previous and current issues faced by textual document clustering algorithms to help text domain researchers understand these issues. This study provides researchers and students an overview about textual document clustering algorithms. Furthermore, this study can encourage researchers to find solutions to these issues.

Usually, clustering algorithms consider that document collections are static and are processed as a whole. However, in contexts where data is constantly being produced (e.g. the Web), systems that receive and process documents incrementally are becoming more and more important. We propose OHDOCLUS, an online and hierarchical algorithm for document clustering. OHDOCLUS creates a tree of clusters where documents are classified as soon as they are received. It is based on COBWEB and CLASSIT, two well-known data clustering algorithms that create hierarchies of probabilistic concepts and were seldom applied to text data. An experimental evaluation was conducted with categorized corpora, and the preliminary results confirm the validity of the proposed method.

Over the past few decades, the volume of existing text data increased exponentially. Automatic tools to organize these huge collections of documents are becoming unprecedentedly important. Document clustering is important for organizing automatically documents into clusters. Most of the clustering algorithms process document collections as a whole; however, it is important to process these documents dynamically. This research aims to develop an incremental algorithm of hierarchical document clustering where each document is processed as soon as it is available. The algorithm is based on two well-known data clustering algorithms (COBWEB and CLASSIT), which create hierarchies of probabilistic concepts, and seldom have been applied to text data. The main contribution of this research is a new framework for incremental document clustering, based on extended versions of these algorithms in conjunction with a set of traditional techniques, modified to work in incremental environments.

2012

Fast and high quality document clustering is one of the most important tasks in the modern era of information. With the huge amount of available data and with an aim to creating better quality clusters, scores of algorithms having quality-complexity trade-offs have been proposed. Some of the proposed algorithms attempt to minimize the computational overload in terms of certain criterion functions defined for the whole set of clustering solution. In this paper, we have proposed a novel algorithm for document clustering using a graph based criterion function. Our algorithm is partitioning in nature. Most of the commonly used partitioning clustering algorithms are inflicted with the drawback of trapping into local optimum solutions. However, the algorithm proposed in this paper usually leads to the global optimum solution. Its performance enhances with the increment in the number of clusters. We have carried out sophisticated experiments wherein we have compared our algorithm with two well known document clustering algorithms viz. k-means and k-means++ algorithm. The results so obtained confirm the superiority of our algorithm.

Information Sciences, 2013

Clustering has become an increasingly important and highly complicated research area for targeting useful and relevant information in modern application domains such as the World Wide Web. Recent studies have shown that the most commonly used partitioning-based clustering algorithm, the K-means algorithm, is more suitable for large datasets. However, the K-means algorithm may generate a local optimal clustering. In this paper, we present novel document clustering algorithms based on the Harmony Search (HS) optimization method. By modeling clustering as an optimization problem, we first propose a pure HS based clustering algorithm that finds near-optimal clusters within a reasonable time. Then, harmony clustering is integrated with the K-means algorithm in three ways to achieve better clustering by combining the explorative power of HS with the refining power of the K-means. Contrary to the localized searching property of K-means algorithm, the proposed algorithms perform a globalized search in the entire solution space. Additionally, the proposed algorithms improve K-means by making it less dependent on the initial parameters such as randomly chosen initial cluster centers, therefore, making it more stable. The behavior of the proposed algorithm is theoretically analyzed by modeling its population variance as a Markov chain. We also conduct an empirical study to determine the impacts of various parameters on the quality of clusters and convergence behavior of the algorithms. In the experiments, we apply the proposed algorithms along with K-means and a Genetic Algorithm (GA) based clustering algorithm on five different document datasets. Experimental results reveal that the proposed algorithms can find better clusters and the quality of clusters is comparable based on F-measure, Entropy, Purity, and Average Distance of Documents to the Cluster Centroid (ADDC).

2000

This paper presents the results of an experimental study of some common document clustering techniques. In particular, we compare the two main approaches to document clustering, agglomerative hierarchical clustering and K-means. (For K-means we used a "standard" K-means algorithm and a variant of K-means, "bisecting" K-means.) Hierarchical clustering is often portrayed as the better quality clustering approach, but is limited because of its quadratic time complexity. In contrast, K-means and its variants have a time complexity which is linear in the number of documents, but are thought to produce inferior clusters. Sometimes K-means and agglomerative hierarchical approaches are combined so as to "get the best of both worlds." However, our results indicate that the bisecting K-means technique is better than the standard K-means approach and as good or better than the hierarchical approaches that we tested for a variety of cluster evaluation metrics. We propose an explanation for these results that is based on an analysis of the specifics of the clustering algorithms and the nature of document data.

Data Mining and Knowledge Discovery, 2005

Fast and high-quality document clustering algorithms play an important role in providing intuitive navigation and browsing mechanisms by organizing large amounts of information into a small number of meaningful clusters. In particular, clustering algorithms that build meaningful hierarchies out of large document collections are ideal tools for their interactive visualization and exploration as they provide data-views that are consistent, predictable, and at different levels of granularity. This paper focuses on document clustering algorithms that build such hierarchical solutions and (i) presents a comprehensive study of partitional and agglomerative algorithms that use different criterion functions and merging schemes, and (ii) presents a new class of clustering algorithms called constrained agglomerative algorithms, which combine features from both partitional and agglomerative approaches that allows them to reduce the early-stage errors made by agglomerative methods and hence improve the quality of clustering solutions. The experimental evaluation shows that, contrary to the common belief, partitional algorithms always lead to better solutions than agglomerative algorithms; making them ideal for clustering large document collections due to not only their relatively low computational requirements, but also higher clustering quality. Furthermore, the constrained agglomerative methods consistently lead to better solutions than agglomerative methods alone and for many cases they outperform partitional methods, as well.

2017 13th International Computer Engineering Conference (ICENCO), 2017

K-means algorithm is a well-known clustering algorithm due to its simplicity. Unfortunately, the output of k-means depends on the initialization of cluster centroids. In this paper, we propose a new hybrid approach for document clustering which uses the outputs of single pass clustering (SPC) as an initialization for k-means algorithm. We aim to get the advantages of careful seeding with single pass clustering and the benefits of k-means algorithm. The experimental results state that the proposed approach outperforms traditional k-means algorithm in both unsupervised and supervised evaluation measures especially when the number of required clusters is increased.