Document Clustering Using Semantic Kernels Based on Term-Term Correlations
Authors:
- Ahmed K. FarahatAFAhmed K. Farahat
- Mohamed S. KamelMKMohamed S. Kamel
- 1 December 2009
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 23759232,p. 459-464
Abstract
Document clustering algorithms usually use vector space model (VSM) as their underlying model for document representation. VSM assumes that terms are independent and accordingly ignores any semantic relations between them. This results in mapping documents to a space where the proximity between document vectors does not reflect their true semantic similarity. In this paper, we propose the use of semantic kernels that are based on term-term correlations for improving the effectiveness of document clustering algorithms. The used kernels measure proximity between documents based on how their terms are statistically correlated. We analyze semantic kernels that capture different aspects of correlations between terms, and evaluate them by conducting experiments on different benchmark data sets. Results show that the proposed method achieves significant improvement in document clustering compared to VSM.Keywords
This publication has 14 references indexed in Scilit:
- Clustering Documents Using a Wikipedia-Based Concept RepresentationLecture Notes in Computer Science, 2009
- A tutorial on spectral clusteringStatistics and Computing, 2007
- Kernel Methods for Pattern Analysis2004
- Empirical and Theoretical Comparisons of Selected Criterion Functions for Document ClusteringMachine Learning, 2004
- Document clustering based on non-negative matrix factorization2003
- Co-clustering documents and words using bipartite spectral graph partitioning2001
- Concept Decompositions for Large Sparse Text Data Using ClusteringMachine Learning, 2001
- Document clustering using word clusters via the information bottleneck method2000
- Projections for efficient document clusteringACM SIGIR Forum, 1997
- Generalized vector spaces model in information retrieval1985