Clustering by Enhancing Co-occurrence Frequency

摘要

Despite that the methodologies for document clustering are abundant in the research literature, the need for it still persists. Due to the large dimensionality and huge amount of resources to be searched, there is tremendous need for algorithms to be developed to mine the knowledge. In this research, the documents are represented using the term frequency, and it is converted into a vector by augmenting some of the features using frequent sets. Bisecting k-means algorithm is used to cluster the data. By changing the bounds on the similarity measures, various clusters can be obtained.