一种基于框架结构的专有名词自动识别方法

本文提出了一种基于框架结构的专有名词统一识别方法。该方法首先根据专有名词的成词特点及出现的上下文环境，重新定义语料属性；然后，提出了属性标注点（AP）的概念，对训练语料进行初次标注，并采用错误驱动的学习方法来获取规则；最后，结合规则和实例对文本进行专名识别。实验表明，该方法在测试样本集上准确率最高可以达到
到92．3％，召回率最高可以达到80．4％，是一种有效的专有名词识别方法。

A Framework-Based Approach for the Automatic Identification of Proper Nouns

In this paper, a method based upon the framework structure is proposed to identify proper nouns. First, the properties of corpus are defined according to the characteristics of proper nouns and their contexts. Then the concept of attribute point is put forward, and rules are collected through an error-driven learning algorithm after labeling train corpus for the first time. Finally, rules and instances are assembled together to identify proper nouns in the texts. The results of experiments show that the precision and the recall rate are up to 92. 3% and 80. 4% respectively, which illuminate that our method is effective in identifying proper nouns.