LESK ALGORITHM AND BABELFY SYSTEM FOR DISAMBIGUATION

Disambiguation is a relevant scientific field of research in language theory and natural language processing. Nowadays the task of qualitative removing of ambiguity is still not solved, nevertheless, several approaches to word sense disambiguation are available. The article describes the experiment of testing systems of word sense disambiguation – the Lesk algorithm and Babelfy system. The systems belong to different approaches. The Lesk algorithm runs on the NLTK library and software package and Babelfy is based on the Babelnet semantic network. The testing was conducted using several sentences containing ambiguous words, expressions, phrasal verbs, homonyms and other ambiguous constructions. During the experiment, the quality of the work of the systems was evaluated. According to the statistical analysis of errors it can be concluded, that the quality of work of systems for removing ambiguity is not high enough. In conclusion, the possible causes of errors of disambiguation systems are described and a solution to improve them is proposed.