Ontologically Supported Semantic Matching
Atanas Kiryakov, Kiril Simov · DSpace repository (University of Tartu) · 2000
Evaluation of the closeness of two texts is a subtask for FTR and IR systems. The basic means used to accomplish it is the matching of atomic text entities (ATEs) such as words, stems, simple phrases and/or concepts. We address the question how concepts can be used as ATEs more eciently in order to match \\small duck" with \\small bird". The onto-matching technique introduced in the paper makes extensive use of lexical ontologies similar to WordNet. We work with two tasks in mind: query expansion and text concept indexing. We outline some arguments showing why onto-matching is useful and how it can be implemented. Also, we conducted some experiments with query expansion for AltaVista. 1 Introduction \\A typical information retrieval task is to select documents from a database in response to a user's query, and rank these documents according to relevance." Strzalkowski et al (1998). The relevance must be dened on the basis of the concepts represented in the text and in the query. Usuall...