An Adaptive Information Retrieval System based on Associative Networks

Abstract

In this paper we present a multilingual information retrieval system that provides access to Tourism information by exploiting the intuitiveness of natural language. In particular, we describe the knowledge representation model underlying the information retrieval system. This knowledge representation approach is based on associative networks and allows the definition of semantic relationships between domain-intrinsic information items. The network structure is used to define weighted associations between information items and augments the system with a fuzzy search strategy. This particular search strategy is performed by a constrained spreading activation algorithm that implements information retrieval on associative networks. Strictly speaking, we take the relatedness of terms into account and show, how this fuzzy search strategy yields beneficial results and, moreover, determines highly associated matches to users' queries. Thus, the combination of the associa-tive network and the constrained spreading activation approach constitutes a search algorithm that evaluates the relatedness of terms and, therefore, provides a means for implicit query expansion.