中文摘要：

本文将基于单语语料的检索技术运用到机器翻译中,构建了一个汉英译文推荐系统,解决传统方法双语料库构建代价高昂的问题,同时提高最终译文的流畅性.译文推荐系统包括查询翻译和信息检索两部分：查询翻译根据给定的一组中文,生成,best英文结果;信息检索评价目标语言与候选译文的相似程度.系统综合两部分得分返回推荐译文.考虑到尽best结果与候选译文的词序一致性,采用Levenshtein距离使得排序结果更加合理.在英汉数据集上的实验表明：在不同n阶语言模型下,译文推荐系统都有很好的表现,加入Levenshtein距离取得了最高70.83%的/测度值.

英文摘要：

In this study,we apply a retrieval technology based on a monolingual corpus to machine translation and construct a Chinese-English translation recommendation system. The system solves the problem of conventional ap-proaches that mainly rely on a parallel corpus,which is difficult to collect. It also improves the fluency of the final translation references. The translation recommendation system combines query-translation and information retrieval. For a given set of Chinese queries,the query-translation function generates N-best English results and the iniorma- tion retrieval function computes the similarity of the query and the candidate translation. The two scores are weigh-ted to return recommended translations. Considering the consistency of word order of the N-best results and the translation candidates,we use Levenshtein-distance to obtain more rational retrieval results. Experiments on Eng- lish-Chinese data sets show that,under different n -order language models,the proposed translation recommendation system demonstrates good performance and achieves a maximum f-measure value of 70. 83% using Levenshtein- distance.