A Stastical Machine Translation System Between Mongolian And Chinese

In the modern world, along with the rapid increase of information, more and more frequent international communication, and especially the popularization of Internet, the potential need of machine translation system is quite huge: People wish to get the correct information from one language to another. The development of Machine Translation (MT) system just makes up this vacancy. Since the first experiment made by the Georgetown University in 1954, MT has developed more than 50 years. Through these years of flexuous development of MT system, people have many different attitudes toward it. But, what we cannot neglect is that, with the hard work of researchers MT has significantly grown up in both technique and practical use.However, current MT systems by and large greatly utilize syntax rules written by linguists, and along with more and more complex syntax rules, the translation result does not improve effectively. To solve this limitation, Statistical Machine Translation system (SMT) revives. Although SMT is not proposed to solve all existing problems, it does improve the traditional translation method. In this paper, we discussed the correlative technique of MT and put forward how to establish a SMT between Mongolian and Chinese, combining the method of preprocessing the Mongolian words and expanding the bilingual dictionary, which guarantees the system to provide more reasonable translation results ultimately.Integrating the knowledge and technology of the information retrieval , computational linguistics, data mining etc, we constructed a statistical machine translation system between Mongolian and Chinese. Moreover, by looking into the statistical method and Mongolian language characteristic, we did several amendments to the system. The experiment result indicates that our method is reasonable, and it is worth to further research for Monglian and Chinese machine translation and even the whole information technology field.