Based On Named Entity Uygur-Chinese Translation Rules And Resources Construction Research

With the spread of education, the people of Xinjiang minority gradually increase the level of education. Xinjiang minority daily increase the information media needs. The number of web pages in Uygur language that increased year by year. Xinjiang news and information web site mainly includes political, economic, military, diplomatic and other social public report. People pay most attention to the news from every country and region, we learned that Xinjiang bilingual news media (including the government all kinds of documents, etc.),there are some problems in relation to translate the finance, date, time and so on, the accuracy of translation is low. In the face of massive information, however, to obtain accurate information is not only need the researchers to solve this problem, but also is the demand for government workers and other workers, too. Translate the number of phrases in web news and government documents correctly is an important part in statistical machine translation. For the propose of this, the main research work of this paper is as follows:First:At first, we have collected the Uygur-Chinese bilingual parallel corpora, sorting and processing this corpora. The main source of corpus collection is download from Xinjiang news website.Second:We have detailed classify the numbers, dates and time phrases. Based on the analysis of Uygur-Chinese time and number words in two languages, on the basis of classification and formulated the corresponding rules of recognition and translation respectively.Third:Artificial writing rules about recognition and translation. Time, number, date this kind of words are appear in the corpus, writing rules is the core of this paper.Finally:We use based on the rules method to identify and translate. Translate numbers and time phrase that appears in webpage correctly, to lay the foundation for the improvement of statistical machine translation system in the further.Innovation of this article is that the current domestic influential with Baidu, Google, Youdao online translation system, etc., but they can only translate the majority language translation, and do not implement between minority language and other languages translation, not to mention translate the numbers and time phrase in the Uygur-Chinese language. This article adopts the method of rule-based implements Uygur-Chinese numbers and time phrase translation.In this paper our research and the experimental results show that the method of writing rules can effectively improve the digital time and phrase translation, thus greatly improve the statistic machine translation quality.