Modeling a logical network of relations of semantic items in superphrasal unities

In existing systems of machine translation the quality of automatic translation depends on mistakes of morphological, syntactic and semantic processing. The mistakes are mostly caused by mistakes of semantic processing and in particular by a wrong choice of a translation equivalent of polysemantic words. The article suggests a model of a logical net of relations of translation equivalents of polysemantic words in superphrasal unities. Finite predicates are used to build the network. Every polysemantic lexeme, sentence, idioglossary are designated by means of subject variables. Their meanings are determined by the predicate of object identification by means of a given variable. The logical network under consideration allows defining the meaning of a translation equivalent using the meanings of the previous polysemantic words in superphrasal unities.