摘要
根据纪检监察领域的任务需求,构建了纪检监察事件语料库为数据集,使用BIOES序列标注方法标记该数据集的实体,并提出BiLSTM-CRF(双向长短期记忆网络-条件随机场)深度学习模型进行纪检监察事件的命名实体识别,该方法对事件中纪检监察机构,人名以及该嫌疑人所受处分名三类命名实体进行识别。采用BiLSTM,BiLSTM-CRF进行对比实验。实验结果显示,使用的方法对三类实体识别的P、R、F值分别为99.63%,99.63%,99.63%,验证了所提方法在纪检监察领域的有效性,证明本研究可以有效获取纪检监察事件中的重要实体信息。
According to the task requirements in the field of discipline inspection and supervision, a corpus of discipline inspection and supervision events was constructed as a data set, the entities of the data set were labeled using the BIOES sequence labeling method, and BiLSTM-CRF(Bidirectional Long Short-Term Memory Network-Conditional Random Field) deep learning model was proposed to recognize the named entities of discipline inspection and supervision events. This method identified three types of named entities in the incident, the disciplinary inspection and supervision agency, the name of the person, and the name of the suspect’s punishment. BiLSTM and BiLSTM-CRF were used for comparative experiments. The experimental results show that the P, R and F values of the three types of entities are 99.63%, 99.63%, 99.63% respectively. The validity of the proposed methods in the field of discipline inspection and supervision is verified, and this study can effectively obtain important entity information in the disciplinary inspection and supervision events.
作者
樊昊
陈俊杰
高静
刘晓玲
FAN Hao;CHEN Jun-jie;GAO Jing;LIU Xiao-ling(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot Inner Mongolia 010000,China;Inner Mongolia Autonomous Region Key Laboratory of Large Data Research and Application of Agriculture and Animal Husbandry,Hohhot Inner Mongolia 010000,China)
出处
《计算机仿真》
北大核心
2022年第6期464-468,共5页
Computer Simulation
基金
国家自然科学基金项目(62066037)
内蒙古自治区科技计划项目(2019GG372)
内蒙古纪检监察大数据实验室开放课题基金(IMDBD2020016)。
