A NEW TERM WEIGHTING SCHEME FOR DOCUMENT CLUSTERING
A. Keerthiram Murugesan, Jun Zhang · 2011
Abstract — In this paper, we present a Cluster-Based Term weighting scheme (CBT) for document clustering algorithms based on Term Frequency- Inverse Document Frequency (T F − IDF). Our method assigns the term weights using the information obtained from the generated clusters and the collection. It identifies the terms that are specific to each cluster and increases their term weight based on their importance. We used the K-means partitional clustering algorithm to compare our method with three widely used term weighting schemes such as Norm − T F, T F − IDF, and T F − IDF − ICF. Our experimental results show that the new method outweighs the existing term weighting schemes and improves the result of a clustering algorithm.