РАЗГРАНИЧЕНИЕ ГЛАГОЛЬНОЙ СИНОНИМИИ И ТРОПОНИМИИ ПРИ ПОСТРОЕНИИ ЭЛЕКТРОННОГО ТЕЗАУРУСА

The article examines the construction of a semantic classification of Russian verbs that is based on a clear distinction between hypo-hypernymic and synonymic relations. Such classifications are used to create electronic thesauri, and the first experience of compiling this kind of dictionaries is the WordNet database. Horizontal semantic relations are represented by synonymy; they group lexical units in sets called synsets. Vertical genus-species (or hypo-hypernymic) relations between verb synsets form the field of troponymy. It should be noted that lexicography has no clear criteria of synset formation; as a result, hypo-hypernyms often appear in synonym sets. In rigidly structured electronic thesauri such blending should be avoided. Using the component analysis, step-by-step identification, linguistic test and existing dictionaries, we studied verbs from the semantic group of opening something. First of all, we identified troponymic relations and constructed chains of verbs from the most concrete to the abstract ones. Then we formed verb synsets. Some cases of blending synonymy with troponymy in the existing dictionaries were revealed. In spite of the fact that in speech hypo-hypernymic pairs are often used to express the same meaning, in text processing programs such a substitution can result in partial distortion of meaning. To avoid this kind of errors, we have determined a conventional boundary beyond which hypo-hypernyms cannot substitute each other in certain contexts. This boundary can be determined by applying the stepby-step identification method if two verbs enter into relations of mutual definition.