基于词向量的句子相似度计算及其应用研究_郭胜国1

现代电子技术

Modern Electronics Technique

2016 年 7 月 1 日

第 39卷第 13 期

Jul. 2016

Vol. 39 No. 13

doi：10.16652/j.issn.1004⁃373x.2016.13.024

近年来，机器翻译的需求量越来越大。从学术文献

。

计算机的翻译结果仅仅作为一个辅助性的参考，最

终的结果由使用者根据候选结果的好坏进行修改取舍，

实例语料库可以结合具体领域进行分门别类的应用，使

得其翻译效果进一步提升，从而更加契合用户的需要。

由于领域性的差异，尤其是专业术语较多的时候，这一

优点在特定领域的翻译中显得更加突出

1 计算机辅助翻译及句子相似度计算

相似度不仅受到共现词汇数量的影响，同时还要结

基于词向量的句子相似度计算及其应用研究

（郑州财经学院 计算机系，河南 郑州 450044）

GUO Shengguo，XING Dandan

For the specific field，the training corpus about the field was collected，and the word vector model of English and Chinese was

constructed by using word2vec developed by Google to design and implement the similarity calculation method of Chinese sen⁃

tences. The similarity calculation method of combining word vector Jaccard similarity with word vector dependent syntax is pro⁃

posed. The experiment results show that effect of the proposed method has more improvement than that of the traditional method.

The similarity algorithm of relevant English and Chinese sentences is packaged by means of interface form. The similarity calcu⁃

lation module of Huajian IAT system was put into practical application as a commodity software.