Alexis COLLOMB, Kozo OI, Eiichiro SUMITA
Construction and Preliminary Experiments
of Document Retrieval System
Based on Similarity Between Words
Abstract:In this report, we present an Information Retrieval System combining three types of computation: a semantic distance calculation using a thesaurus, a classical term-weighting method and a physical distance computation intended to reflect how scattered query words can be found in documents. We present our experiments and argue about the quality of the results. Eventually, we introduce new directions and attempt to define heuristic orientations for applying semantics to Information Retrieval.