Finding hierarchical structures of document collections By using tolerance relations

摘要

We develop a hierarchical clustering algorithm based on Tolerance Rough Set Model (TRSM). Text clustering is one of ways to find the structure of the text collection. The quality of text clustering depends not only on the clustering algorithm but also on the document representation model. We aim to enrich representations concerning with documents and their distance according to semantic relations introduced by TRSM. The model offers a way of considering semantics relatedness between documents. It is an extension of the equivalence rough set model by employing tolerance relations instead of equivalence relations. The main advantages of the proposed model are it is more appropriate for textual data and the computation can be done efficiently. Based on the tolerance rough set model, we develop a hierarchical document clustering algorithm. The algorithm is evaluated and validated experimentally on test collections. The results suggest that this clustering algorithm can be well adapted to text mining.