一种新的自纠错句对齐算法的研究与实现

本文提出了一种基于词性的判别函数,用来自动识别句子对齐结果的正确性。通过反向采用基于长度的对齐算法,纠正了初步对齐产生的连续性错误。该方法节省了人力,同时提高了对齐结果的正确率。实验结果表明,本模型开放集测试的召回率达到97.60%,已满足实际应用的需求。

A New Self- Correction Algorithm for Sentence Alignment

The paper presents the POS decision function to rec-ognize the wrong results automatically. And to the successionalwrong results, it corrects the error with the reverse algorithmbased length .The algorithm saves the manpower and improvesthe precision level. Experiment results illustrate that the perfor-mance is excellent, and the recall rate is 97.60%.