维吾尔语-汉语量词短语的神经机器翻译  

Neural machine translation of Uyghur-Chinese quantifier

作　　者：阿依古丽·哈力克 卡哈尔江·阿比的热西提[1,2] 艾山·吾买尔[1,2] 吐尔根·依布拉音[1,2] Ayiguli Halike;Kahaerjiang Abiderexiti;Aishan Wumaier;Tuergen Yibulayin(School of Information Science and Engineering,Xinjiang University,Urumqi 830046,China;Xinjiang Key Laboratory of Multi-Language Information Technology,Xinjiang University,Urumqi 830046,China)

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046 [2]新疆大学新疆多语种信息技术重点实验室,新疆乌鲁木齐830046

出　　处：《计算机工程与设计》2019年第9期2649-2653,共5页Computer Engineering and Design

基　　金：国家重点基础研究发展计划基金项目(2017YFB1002103);国家自然科学基金项目(61462083、61762084、61463048、61262060)

摘　　要：针对语料不足引起的维吾尔语-汉语量词短语翻译性能不佳及歧义量词的翻译结果乱等问题,提出基于多编码器多解码器的维吾尔语-汉语量词短语神经机器翻译方法。将汉语语料进行自动处理和筛选,收集包含量词的维吾尔语-汉语量词短语语料,使用多编码器多解码器结构,搭建具有更强生成能力的多编码器多解码器维汉神经机器翻译模型。计算BLEU值得到了36.41,并通过对比实验结果,验证了所提方法的有效性。To solve the problems of poor performance of Uyghur-Chinese quantifier translation and the confusion of translation results of ambiguous quantifiers caused by insufficient corpus,the neural machine translation method based on multi-encoder multi-decoder for Uyghur-Chinese quantifier was proposed.The Chinese corpus was automatically processed and filtered,and the Uyghur-Chinese quantifier corpus containing quantifiers was collected.The neural machine translation model based on multi-encoder multi-decoder for Uyghur-Chinese was constructed using the multi-encoder multi-decoder structure.The BLEU value calculated is 36.41.By comparing the experimental results,the effectiveness of the proposed method is verified.

关 键 词：维汉神经机器翻译 歧义量词 神经网络 多编码器 多解码器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]