본문 바로가기
  • Home

Semantic Classification of Lexical Translation Patterns Extracted from Bilingual Corpus and Application of Lexical Translation Patterns to MT System

Sung-Kwon Choi 1 Youngkil Kim 1

1한국전자통신연구원

Accredited

ABSTRACT

We could make the results translated by a machine translation system natural if we make an implicit translation competence of professional translator explicit and apply it to the machine translation system. The lexical translation pattern can be taken as the explicit translation competence of professional translator. The purpose of this paper is to describe semi-automatic construction and semantic classification of the lexical translation patterns that make the machine translation results natural. The lexical translation patterns can be built semi-automatically by two ways. One is to extract automatically the idiomatic expressions of source language from bilingual corpus and construct the lexical translation pattern manually. The other way is to construct manually the lexical translation pattern by comparing the translation result of professional translator with the result translated by machine translation system. We are able to classify the lexical translation patterns into two semantic groups. One group is related to the idiomatic lexical translation patterns. The other group contains the lexical translation patterns with non-animate subject phrase, quantifier phrase, passive phrase, and verb-derived noun phrase.

Citation status

* References for papers published after 2022 are currently being built.