Machine translation system modeling based on sentences comparison

Abstract

The problem urgency is determined by an increasing need for natural language processing systems especially in the area of construction and civil engineering due to an enormous amount of document in foreign languages to deal with. Author’s purpose is to figure out an approach to machine translation system based on initial sentence and online dictionaries examples comparison. Although online dictionaries are designed to be used by humans, application of some principles based on the theory of graphs and syntactic analysis as well as methods of sets allows to use them as thesaurus for natural processing systems. The author proves the reliability of this approach by using the properties of a relation such as reflexivity, symmetry and transitivity. As a result the author displays a part of a syntactic analysis block of his computer program as well as a block of the program responsible for comparing sentences on the basis of fundamental products. At the end of the paper the author comes to the conclusion that such an approach can be extended not only to the translation problem but also to a broad range of natural language processing issues.