基于CRF与规则相结合的中文电子病历命名实体识别研究

A study on the named entity recognition of Chinese electronic medical record based on combination of CRF and rules

目的:探讨基于条件随机场(conditional random field,CRF)与规则相结合的中文电子病历命名实体识别。方法:基于条件随机场和规则相结合的方法来识别实体,将语言、关键词、词典等作为特征,识别出的结果再利用规则进行优化。结果:与条件随机场的方法相比,条件随机场和规则相结合的方法识别准确率提高到78.98%,召回率和F值也提高到88.37%和83.41%。结论:基于条件随机场和规则相结合的方法来识别实体,准确率和召回率满足应用需求,为电子病历后续研究奠定了基础。

Objective：To explore the named entity recognition of Chinese electronic medical record based on the combination of conditional random field （CRF） and rules. Methods：Entities are recognized based on the combination of conditional random field and rules. Language, keywords, dictionaries are used as recognition features, and the recognition results are optimized by the rules. Results：Compared with the method of conditional random field, the accuracy of the method combining the conditional random field with the rules is improved to 78.98%, and the recall rate and F value are also increased to 88.37% and 83.41%. Conclusion： The accuracy and recall rate based on the method combining the conditional random field with the rule to identify entities can meet the apphcation requirements, which will lay the foundation for the follow - up study of electronic medical record.