Possibilities of computer-aided extraction of hyper-hyponymic pairs from dictionary definitions of verbs

The paper concerns computer-aided methods for extraction of semantic relations between lexical units. The aim of the research is to elaborate a method of genus-species relations extraction from dictionary definitions. The suggested method is based on the idea of lexico-syntactic patterns and its adaptation to verbal vocabulary. Definitions from six dictionaries are analyzed, probable genus-species pairs are extracted from the definitions with the help of previously defined lexical markers. The result of automatic processing is verified manually. The results of the method application and the most frequent mistakes made during computer processing are analyzed and possible ways of method improvement are suggested. The primary approbation of the method showed that its further development and elaboration have good prospects, and it can be a base for creation of an automatic genus-species relations extraction software showing high accuracy. The results of the research can be used in different domains of computational linguistics and during further research in theoretical semantics.