An Hybrid Approach for Improving Word Sense Disambiguation and Text Clustering

Abstract— In this paper we suggest a new approach to represent text document collections, integrating background knowledge to improve clustering effectiveness. Background knowledge is inferred from previous classification tasks on the same collection and it’s used as context to assign semantic values to words. The WordNet ontology acts as a source repository of semantic meanings. More specifically we propose a new approach to Word Semantic Disambiguation, based on the evaluation of functions related with syntactical and statistical properties of the WordNet synsets or with contextual and synonymical informations.

• Type: Conference
• Year: 2007
• Where: IRCDL
• Authors: Paolo Casoto, Carlo Tasso
• Keywords: Background Knowledge, Digital Library, IRCDL 2007, Semantic, Word Semantic Disambiguation