基于条件随机场的中文人名识别

Recognition of Chinese Person-Name Based on Conditional Random Fields

中文人名数量众多,规律各异,使得中文人名识别成为中文分词的重点和难点,中文人名识别技术的突破将对提高汉语自动分词和句法分析的准确性有着很重要的作用。本文根据中文人名的特点,利用条件随机场模型进行人名识别,通过第四届Bakeoff评测语料上测试,召回率为80.97%,准确率为94.42%,综合指标F为87.18%。

Now,recognition of person names becomes a difficult and key point in Chinese word segmentation.This technology improvement is very important to Chinese word segmentation and syntactic study.This paper implements a name recognition test based on conditional random fields.In the test,the recall,precision and F-measure has reached 80.97%,94.42% and 87.18%,respectively.