The Methods of Estimation of the Degree of Similarity of Sentences in a Natural Language Based on the Link Grammar

Abstract

The paper describes the methods of comparison of the sentences in a natural language for estimation of their similarity. To solve this problem, it is possible to use the semantic-syntactical relations between words constructed by the software system Link Grammar Parser. The results of our research are planned to be used in information retrieval systems. The application of the methods here considered to studies of Turkic languages is briefly described.