基于条件随机场的中文文本弱定位地名识别

Chinese Text Weak Positioning Placename Recognition Based on Conditional Random Field

针对中文文本,通过引入语言学相关知识,分析了强定位地名和弱定位地名的用词特征和上下文特征,提取形式化的规则。此外,提出一种基于条件随机场的弱定位地名识别方法,将弱定位地名识别问题转化为序列标注问题。以150篇共18万字的新浪网新闻页面文本为例进行实验验证,结果表明,提出的方法能够有效识别弱定位地名,召回率为90.57%,准确率为92.36%,F值为91.46%。

By introducing linguistic knowledge,the author analyzed the word features and context features of strong and weak location names of Chinese text,and extracted formal rules.In addition,a weak location name recognition method based on conditional random field was proposed,which transformed the problem of weak location name recognition into sequence labeling problem.An experimental verification was carried out by taking 150 Sina news pages with 180000 words as an example.The results show that the proposed method can effectively identify weak location names,and the recall rate is 90.57%,the accuracy rate is 92.36%,and the F value is 91.46%.