铁路客站安全风险事件知识图谱构建及应用  

Construction and application of knowledge graph of railway passenger station safety risk events

作　　者：白伟 王小书 张煜山 杨国元 BAI Wei;WANG Xiaoshu;ZHANG Yushan;YANG Guoyuan(Institute of Computing Technologies,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China)

机构地区：[1]中国铁道科学研究院集团有限公司电子计算技术研究所,北京100081

出　　处：《铁路计算机应用》2025年第3期1-6,共6页Railway Computer Application

基　　金：中国铁道科学研究院集团有限公司科研项目(2023YJ125)。

摘　　要：铁路客运车站(简称:客站)安全风险事件数据多以文本形式进行存储,难以高效、快速查询。为充分发挥数据价值,文章研究了铁路客站安全风险事件领域知识图谱的构建及应用。提出了适用于铁路客站安全风险事件管理的知识图谱构建框架;研究了基于BERT-BiLSTM-CRF模型的知识抽取方法,并以某客站安全风险事件数据为基础进行数据层构建,试验表明该模型效果优于其他主流识别技术;构建了面向铁路客站安全风险事件知识图谱,并通过Neo4j实现图数据的结构化存储和展示;设计了基于该知识图谱的安全风险事件智能问答系统,该系统能够针对用户所提问题,提供满足真实场景与需求的高效、智能化应答,有效提高铁路客站安全风险事件的检索效率。The safety risk event data of railway passenger stations are mostly stored in text form,which is difficult to efficiently and quickly query.To fully leverage the value of data,this paper investigated the construction and application of a knowledge graph of railway passenger station safety risk events,proposed a knowledge graph construction framework suitable for railway passenger station safety risk event management,studied the knowledge extraction method based on BERT BiLSTM CRF model,and constructed the data layer based on the safety risk event data of a certain passenger station.The experiment showed that the model had better performance than other mainstream recognition technologies.The paper constructed a knowledge graph for railway passenger station safety risk events,implemented structured storage and display of graph data through Neo4j,and designed safety risk event intelligent question answering system based on this knowledge graph,which can provide efficient and intelligent responses to user questions that meet real scenarios and needs,effectively improve the retrieval efficiency of railway passenger station safety risk events.

关 键 词：安全风险事件 知识图谱 BERT-BiLSTM-CRF Neo4j 智能问答 

分 类 号：U298[交通运输工程—交通运输规划与管理] U291.6[交通运输工程—道路与铁道工程] TP39[自动化与计算机技术—计算机应用技术]