The Research Of Tibetan-Chinese Online Machine Translation Based On The Phrase

Tibetan-Chinese online machine translation is a hotspot and difficulty in the field of Tibetan natural language processing, Tibetan-Chinese machine translation is put Tibetan automatic translation into Chinese by computer, the process of translation is the translator for Chinese translation can be done via the Internet. At present, based on the statistics of Tibetan-Chinese machine translation method in a leading position, among them, the Tibetan-Chinese machine translation method based on the phrase as a whole is better than that of the machine translation method based on word.Based on statistical machine translation of the relevant theory and Tibetan and working principle of the machine translation system research, utilization of Moses, IRSTLM, GIZA++, XMLRPC open source toolkit built a Tibetan online translation system based on the phrase. Among them, the phrase is not in the traditional sense of the phrase, it refers to a series of word sequences. The research content includes:1 using open source tools IRSTLM completed the training of Chinese language model, using Moses, GIZA++ word alignment tools to complete the training of Tibetan translation model, finally through Moses completed the construction of the entire Chinese machine translation system.2 using open source tools XMLRPC realized the function of Chinese online translation. Set up by the Tibetan and online translation system can start in the form of server, the translator can through the web page to access the system, make the system have the function of online translation.3 For tibetan-chinese bilingual corpus in order to solve the limitation problem caused by sparse data, to set up by the Tibetan and online translation system as the research platform, adopt the method of the extended dictionary manually expanded about 1550 words, to a certain extent, improve the quality of the Tibetan and machine translation.4 using BLEU automatic metering technology, hide into the extended dictionary before han online translation system for automatic testing. Come into the extended dictionary information before the BLEU value of 0.287361, after extended dictionary information with BLEU value of 0.294997.May safely draw the conclusion:from the result of experiment evaluation into the extended dictionary information can to a certain extent, improve the quality of Tibetan translation machine translation.