摘要
为了快速识别高速公路风险车辆,采用知识图谱推理的识别方法,研究了高速公路运行车辆多元历史信息与安全风险之间的潜在联系。首先,确定了风险车辆类型和文本数据的来源,建立了考虑一对多、多对一和多对多关系的表示学习模型;然后,以最小化对数损失为目标,基于开放世界假设进行了样本训练;最后,采用随机游走推理模型进行知识推理,构建了高速公路风险车辆知识图谱。结果表明:建立的模型能兼顾准确度和计算效率,有效识别高速公路运行的潜在风险;语义关系识别领域的知识图谱技术可应用于高速公路运行风险预警管控。
In order to quickly identify highway risk vehicles,the potential relationship between multiple historical information of highway vehicles and safety risk was studied by using the identification method of knowledge graph reasoning.Firstly,the types of risk vehicles and the sources of text data were determined,and a representation learning model considering 1-to-N,N-to-1 and N-to-N relationships was established.Then,aiming at minimizing the logarithmic loss,the sample training was carried out based on the open world hypothesis.Finally,the random walk reasoning model was used for knowledge reasoning,and the knowledge graph of highway risk vehicles was constructed.The results show that the proposed model can balance the accuracy and efficiency,and effectively identify the potential risks of expressway operation.The knowledge graph technology in the field of semantic relationship recognition can be applied to the early warning and control of highway operation risk.
作者
俞山川
谢耀华
陈晨
周健
YU Shanchuan;XIE Yaohua;CHEN Chen;ZHOU Jian(Research and Development Centre of Transport Industry of Self-driving Technology,China Merchants Chongqing Communications Research&Design Institute Co.,Ltd.,Chongqing 400067,China)
出处
《重庆交通大学学报（自然科学版）》
CAS
CSCD
北大核心
2022年第1期22-28,75,共8页
Journal of Chongqing Jiaotong University(Natural Science)
基金
国家自然科学基金青年科学基金项目(71901190)
广西科技计划项目(桂科AB20159032)。
关键词
交通工程
运行风险
知识图谱
知识推理
表示学习
traffic engineering
operational risk
knowledge graph
knowledge reasoning
representation learning
