基于Naive Bayes的中文人名识别研究

在传统的只统计人名用字的Naive Bayes分类算法的基础上,将人名上下文边界融入其中,并利用从大规模语料库中统计的人名用字、边界模板频率对人名定界,再通过扩散操作召回遗漏人名。该方法简单易行,并能取得很好的效果。实验结果表明,其F值达到了93.28%。

Chinese person name recognition based on Naive Bayes

On the basis of the traditional Naive Bayesian classification algorithm that just considered character of Chinese person name, we brought person name's up and down boundary words in it. In order to overcome the difficulty of boundary defining, we counted Chinese name's character frequency and boundary templates' frequency from tagged corpus. Then these recognized person names are used to match the missed occurrence in the text. The method is easy and the final result is good. Experimental results show that the F-value for recognition of Chinese person name was increased.