Research and design of Chinese-English corpus alignment

Sentence alignment is most important for Chinese-English bilingual corpus alignment. This paper analyzes the features of the length-based method and the lexical-information method. First, the N-Path method is proposed which combines length and lexical-information. This method avoids the insufficiency of length-based method and limitations of lexical-information methods. Secondly, we can get expanded result aggregate and improve the alignment effect using the N-Path alignment method. Finally, we can confirm the best one from the result aggregate by using synonym-expansion alignment. The experiment result shows that our methods can be applied to general text processing and improves alignment accuracy. © 2014 Taylor & Francis Group, London