科研管理

科研管理 ›› 2018, Vol. 39 ›› Issue (7): 131-138.

• 论文 • 上一篇    下一篇

海量食品安全事件下的命名实体识别研究

徐飞1,2,宋英华1,2   

  1. 1.武汉理工大学 中国应急管理研究中心,湖北 武汉430070; 2.武汉理工大学 管理学院,湖北 武汉430070
  • 收稿日期:2017-11-02 修回日期:2018-05-18 出版日期:2018-07-20 发布日期:2018-11-06
  • 通讯作者: 宋英华
  • 基金资助:

    国家社会科学基金重大项目：“基于情报流知识库的我国食品安全技术支撑体系优化策略研究”（15ZDB168）。

A research on identification of the named entity for large-scale food safety incidents

Xu Fei1,2, Song Yinghua1,2   

  1. 1. China Research Center for Emergency Management, Wuhan University of Technology, Wuhan 430070, Hubei, China;
    2. School of Management, Wuhan University of Technology, Wuhan 430070, Hubei, China
  • Received:2017-11-02 Revised:2018-05-18 Online:2018-07-20 Published:2018-11-06

摘要： 对食品安全事件当中的实体进行分析和识别，不仅有助于人们加深对食品安全事件的了解，而且有利于管理者应对食品安全事件。以食品安全事件的新闻报道文本为语料，通过系统地统计和分析人名和机构名的内部与外部特征，在制定的含有多个特征的识别模板的基础上，基于条件随机场模型，本文完成了对食品安全事件当中的机构名和人名这两个命名实体进行识别的任务。通过与最大熵模型的测试结果进行比较，实验表明条件随机场模型的整体性能比较突出，取得了较好的准确率和召回率。

关键词: 条件随机场模型, 特征分析, 实体识别, 食品安全事件

Abstract: It is not only helpful for people who can get a deeper understanding of food safety incident but also beneficial for managers who deal with food safety incidents to analyze and identify the entity of food safety incidents. With the news reports of food safety incident texts as the corpus and by counting and analyzing the internal and external characteristics of the entities of organization name and the person names, the identification task of the entities of organization name and the person names in food safety incidents based on conditional random field model is completed in formulating identification template containing multiple characteristics. The overall performance of the conditional random field model is outstanding in the experimental results comparing with the test results of maximum entropy model, and the accuracy and recall rate is very well.

Key words: conditional random fields, feature analysis, entity identification, food safety incident