RISING OF THE TEXT DOCUMENTS SEARCH PRECISION BY USING THE ADAPTIVE ONTOLOGY
Conceptual graphs are an effective tool for representation of the semantic content of text documents and domain ontology as well. In this article the new method of evaluation of text documents content similarity is proposed. The method consists in representation compared texts as its weighted conceptual graphs supplemented by related context from domain ontology and estimation of a distance between semantic weights centers of these graphs. It is shown that the method satisfies axioms of a metric. Procedures of the automatic tuning of ontology to the specified domain and information needs of user are developed. The results of experiment shows that the taking into account semantics of the used concepts, assertions and significance coefficients from adaptive ontology during the text processing rises the search precision on average 20 %.