中文组织机构名称跟简称的识别.pdfVIP

中文组织机构名称与简称的识别 1 1 1 2 2 沈嘉懿 ，李芳 ，徐飞玉 ，Hans Uszkoreit 1 上海交通大学计算机系 上海，200240 2 德国人工智能研究中心语言技术实验室 摘要：本文提出了一种基于规则识别中文组织机构名称和简称的方法。中文组织机构全称 的识别是通过确定其左右边界实现的:首先借助机构称谓词库获得组织机构的右边界，然后 运用规则匹配以及贝叶斯概率模型决策出最优规则，确定组织机构名称的左边界。对应每一 条全称规则，应用其对应的简称规则，获得该机构全称对应的候选简称。在封闭性测试中， 该方法的总体查全率为88.72%，查准率为92.68%，简称的查全率为73.33%，查准率为77.19%； 在开放性测试中，该方法的总体查全率为 88.8%，查准率为 83.03%，简称的查全率为 67.18%， 查准率为74.14%。该方法实现的系统可以独立运行，也可以嵌入在GATE环境，目前，该方 法已应用在中文关系的抽取系统中，得到了较好的结果。 关键词：组织机构名称识别，组织机构简称识别，词性标注，规则匹配，贝叶斯概率模型 中图分类号：TP391.1 Recognition of Chinese Organization Names and Abbreviations 1 1 2 2 Shen Jiayi ，Li Fang , Feiyu Xu , Hans Uszkoreit 1Department of Computer Science and Technology, Shanghai JiaoTong University 2 German Research Center for Artificial Intelligence Abstract:This paper proposes a method for recognizing Chinese organization names and their abbreviations based on rules. The recognition of Chinese organization names needs to identify the left and right boundary. The right boundary of the organization is identified with the help of organization postfix lexicon. The left boundary is recognized by the optimum rule based on Bayesian probability model. After idendifying an organization name, we can get candidate abbreviations based on abbreviation rules accordingly. In close test, we get the recall of 88.72%, the precision of 92.68% in general and the recall of 73.33%, the precision of 77.19% in abbreviation recognition. In open test, we get the recall of 88.8%, the precision of 83.03% in general and the recall of 67.18%, the precision of 74.14% in abbreviation recognition. There are two implementations, one is stand alone,