A Comparative Study of Clustering versus Classification over Reuter's Collection

People have plenty of information at their disposal. The problem is that, even with the advent of search engines, it is still complex to analyze, understand and select relevant information. In this sense, clustering techniques sound very promising, grouping related information in an organized way. This paper address some problems of the existing document clustering techniques and present the “best star” algorithm, which can be used to group and understand chunks of information and find the most relevant ones.