Knowledge engineering

Semantic text retrieval based on fuzzy set theory

L. A. Pankova, V. A. Pronina

V. A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences, Moscow

Abstract: The crisp models of text retrieval do not take into account fuzziness of information.The fuzzy set theory provides means of handling fuzzy information. In existing works the theory of fuzzy sets is mainly used to represent ontologies and to introduce more flexible ways of formulating queries. In practice, text search models are still based on other approaches. In this area, the fuzzy set theory can offer elegant and intuitively attractive methods. In the paper the concepts of text retrieval are interpreted in terms of the fuzzy set theory. The models of text retrieval on the basis of the fuzzy set theory are proposed. The proposed models are based on the generalization principle and are intuitively simple. The generalization principle is the universal principle of the fuzzy set theory. In these models, the principle of generalization is used to move from relation between concepts towards relation between document and query, namely from relatedness of concepts towards the relevance of document to query. The paper presents the example that shows results of modeling.

Keywords: text retrieval, relatedness, relevance, fuzzy set theory, fuzzy relation, generalization principle.

Bibliographic databases:

•