On Semantic Approaches To Machine Translation

Beginning in the 1950s with attempts to use computers for machine translation (MT), natural language processing (NLP) has become a significant research area. It has practical utility in such domains as man-machine interfaces, intelligent access to databases, computer assisted instruction and machine translation. Our concern in this thesis is NLP from the viewpoint of semantic approaches to problems in MT. The technology of semantic analysis in MT is far from satisfying compared with the technology of syntactic processing. The thesis can be divided into three chapters. In the first chapter, natural language processing is introduced. The second chapter consists of three parts, that is, different approaches to NLP, linguistic components in NLP and theories of semantic analysis. In the third chapter, the semantic issue is addressed and semantic approaches to the problems are introduced and analyzed. To sum up, by analyzing semantic problems in MT and studying on different theories and methods of semantic analysis, a conclusion can be drawn that the trend of developing MT is to integrate rule-based and corpus-based approaches to solve semantic problems in MT. The semantic analysis processing is a critical problem not only in MT, but also in other NLP systems.