基于深度学习的业务稽查规则知识图谱构建  

Construction of knowledge graph of business audit rules based on deep learning

作　　者：余安国 刘继鹏 郭伟 孙志杰 张艳丽 YU Anguo;LIU Jipeng;GUO Wei;SUN Zhijie;ZHANG Yanli(Marketing Service Center(Fund Intensive Control Center and Metrology Center),State Grid Jibei Electric Power Co.,Ltd.,Beijing 100032,China)

机构地区：[1]国网冀北电力有限公司营销服务中心(资金集约中心、计量中心),北京100032

出　　处：《电子设计工程》2022年第7期156-159,164,共5页Electronic Design Engineering

摘　　要：目前的业务稽查规则知识图谱构建方法信息分析能力较差,导致业务稽查规则知识图谱关系识别准确度较低。为了解决上述问题,基于深度学习研究了一种新的业务稽查规则知识图谱构建方法。利用互联网资源对业务规则内容和关键词进行注释和特征描述,将描述内容导入知识图谱数据库中。采用NLP自然语言处理及深度学习模型算法,利用序列标注技术完成知识的识别和提取。利用Bert语义消歧和知识链接技术完成知识融合,构建业务稽查规则知识图谱。实验结果表明,基于深度学习的业务稽查规则知识图谱构建方法能够有效提高信息分析能力,增强识别准确率。At present,the construction method of business audit rule knowledge map has poor information analysis ability,which leads to the low accuracy of business audit rule knowledge map relationship recognition.In order to solve the above problems,a new knowledge mapping method of business audit rules is studied based on deep learning.The Internet resources are used to annotate and describe the content and keywords of business rules,and the description content is imported into the knowledge map database.NLP natural language processing and deep learning model algorithm are used,and sequence annotation technology is used to complete knowledge recognition and extraction.The knowledge fusion is completed by using Bert semantic disambiguation and knowledge link technology,and the knowledge map of business audit rules is constructed.The experimental results show that the knowledge map construction method of business audit rules based on deep learning can effectively improve the information analysis ability and enhance the recognition accuracy.

关 键 词：深度学习 业务稽查 规则知识 图谱构建 

分 类 号：TN391[电子电信—物理电子学]