Research On Word Alignment And Its Use On Automatical Evaluation Technology Of Translation Quality

In the 21st century, with the rapid development of Internet infrastructure and information industry, the course of world economy integration faces unprecedented acceleration, and people from all over the world contact more and more with each other nowadays. In this trend, people become more desired on the acquisition of information and knowledge expressed in one’s non-mother tongue. Accordingly, Multi-Language Information Retrieval, Machine Translation and other related fields become new research focus.Bilingual Word Alignment can be defined as an object for indicating the corresponding words in a parallel sentence pair which have been aligned in the parallel text. It is a basic subject in Machine Translation and is a very useful but difficult research subject in NLP. According to the analysis of current mainstream methods for word alignment through experiments, in the view of various factors, we propose a method that combines heuristic rules and lexicon in this paper. The method makes full use of existing resources, and subsequent application is taken into account. Experiments show that good bilingual word alignment results are achieved in the method based on a small training corpus.Since 1980s, due to certain research achievements in Machine Translation, some practical translation systems came into the world. The research on Machine Translation began to focus on applications. Statistical Machine Translation developed considerably especially in the last 10 years, marking that Machine Translation came into a new stage. As some market-oriented commercial machine translation systems come into the world, translation quality evaluation is increasingly becoming an important issue. At present, translation quality evaluation mainly belongs to manual work. However, manual evaluation has the disadvantages of high cost and low efficiency, which cannot meet the needs of the growing translation market on translation quality evaluation. Therefore, the development of an automatic evaluation system is of great significance for current research.According to the analysis on the procedures of machine translation and the study on the alignment technology in the first part of the research, we propose a translation quality evaluation method based on the alignment technology in this paper. You can easily find pretermissions and other information in translation in terms of alignment information obtained by the alignment technology and make qualitative evaluation without reference translation. Experiments show that use of the alignment technology for translation quality evaluation can relative objectively indicate translation quality.