基于自然语言处理的交通事故隐患关联分析方法

Research on Correlation Analysis Technology of Traffic Accident Hidden Danger Based on Natural Language Processing

随着城市交通管理警务效能的提高,公安交管接处警业务会产生大量的非结构化的文本信息,针对交通事故的分析研判、安全隐患排查等工作,通过人工查询数据库等常规分析方法,难以从大量数据中挖掘出有效的关联关系。对此,本文提出一种基于自然语言处理的交通事故隐患关联分析方法,首先利用自然语言处理技术对非结构化的文本数据进行依存句法分析,抽取事故文本数据中的实体与关系并组建事故事件三元组,然后构建事故因子匹配机制,将三元组的匹配结果转为结构化可编码的数据,结合GPU-Apriori算法挖掘编码后数据的关联关系,最后根据维护的关联结果评价映射关系得出分析结果。

With the improvement of the policing efficiency of urban traffic management, a large amount of unstructured text information will be generated when the public security traffic management takes over the police service.Aiming at the analysis and judgment of traffic accidents and the investigation of potential safety hazards, this paper proposes a traffic management system based on natural language processing. Accident hidden danger association analysis method, firstly uses natural language processing technology to perform dependency syntax analysis on unstructured text data, extracts entities and relationships in accident text data and forms accident event triples, and then converts the matching results of triples into In order to structure and encode the data, the GPU-Apriori algorithm is combined to mine the correlation relationship of the encoded data, and finally the analysis result is obtained by evaluating the mapping relationship according to the maintained correlation result.