A State-of-the-Art Toolkit for Document Clustering

ii Cluster analysis refers to a family of procedures which are fundamentally concerned with automatically arranging data into meaningful groups. These procedures are increasingly being employed in knowledge discovery tasks to assist in the exploration and interpretation of large datasets. Since users may often be unfamiliar with the exact contents of a dataset, clustering can provide a means of introducing some form of organisation to the data, which can also serve to highlight significant patterns and trends. Cluster analysis methods have recently become an important part of commercial and industrial applications for mining data in a variety of domains. In the past, these methods have also been employed to facilitate the discovery of knowledge from large collections of unstructured text. Renewed interest in document clustering has been prompted by the exponential growth in the size of digital document collections, including web pages, e-mail messages and news articles. Another motivating factor has been the increased availability of computing resources, which has encouraged researchers to reconsider the application o