Article · Eastern-European Journal of Enterprise Technologies · February 27, 2017

Development of knowledgeoriented system of machine translation based on the analyticsynthetic text processing

Leonid Lytvynenko, Oleksandr Nikolaievskyi, Valeriy Lakhno, Олена СКЛЯРЕНКО

Abstract

A method for automated syntactic text analysis based on the declarative representation of the rules of syntactic combinability was developed. In this method, in contrast to those existing, the tables of syntactic rules are used not only for the context analysis, but also for defining the subject, predicate, secondary parts of the sentence, as well as superphrase syntactic combinations.A method for software distribution of analytical-synthetic processing of a natural language text in machine translation systems was developed. The developed method, in contrast to the known methods, takes into account conditions of transition to parallel data processing both at the level of processing tasks and depending on the data type.The C# applications, where the developed methods for analytical-synthetic processing of multilingual Russian, Ukrainian and English texts were realized, were implemented by software. It was experimentally proved that the developed software for texts on military subject area allow reducing the number of errors of semantic character by 14–16 % in comparison with the existing machine translation systems through the automated text processing at the level of sign system and the introduction of super-phrase synthesis.

Natural Language Processing TechniquesTranslation Studies and PracticesSyntax, Semantics, Linguistic Variationmachine translationparallel data processingmultilingual text processing

Development of knowledgeoriented system of machine translation based on the analyticsynthetic text processing

Abstract

More by Valerii Lakhno