Text Clustering using One-Mode Projection of Document-Word Bipartite Graphs (2012)

Authors: Ajitesh Srivastava, Axel J. Soto, Evangelos Milios

In this work we have studied the use of one-mode projection of the document-word bipartite graph and the subsequent application of a modularity optimization algorithm to cluster the documents. In particular, we have proposed an alternative and faster algorithm, which works in two-steps: first, finding the documents that are easy to cluster, and then, assigning the remaining documents to the existing or new clusters. We show that the algorithms based on one-mode projections perform significantly better than traditional clustering approaches. In addition, our method has similar or better clustering performance than the most popular algorithm for modularity optimization, while also running four times faster.

Details about this work can be found below:

  1. A. Srivastava, A.J. Soto, E. Milios. “Text Clustering using One-Mode Projection of Document-Word Bipartite Graphs”. 28th ACM Symposium on Applied Computing, 927-932. March 18-22, 2013; Coimbra, Portugal.[paper][bibtex][code]