基于规则推理和贝叶斯网络算法的多方证据关联分析  

Multi Evidence Association Analysis Based on Rule Reasoning and Bayesian Network Algorithm

作　　者：赵晋斌 王凯 李盼 ZHAO Jin-bin;WANG Kai;LI Pan(Unit 61646 of PLA,Beijing 100191,China;Zhongjing Baicheng Technology Co.,Ltd.,Beijing 100000,China;China Justice Big Data Institute Beijing 100043,China)

机构地区：[1]中国人民解放军61646部队,北京100191 [2]中经柏诚科技(北京)有限责任公司,北京100000 [3]中国司法大数据研究院有限公司,北京100043

出　　处：《中国电子科学研究院学报》2022年第5期508-514,共7页Journal of China Academy of Electronics and Information Technology

基　　金：国家重点研发计划资助项(2018YFC0830200,2018YFC0830202)。

摘　　要：目前,中国司法数据存在数据价值密度低,关联性差等问题。从大量的冗杂司法数据中进行文本分类的高效处理,快速提取出有效信息,不仅能够有效地节约法院诉讼服务的人力及空间资源,同时能够为公众随时提供简单、安全、智慧、高效的诉讼智能服务。因此,文中设计出一种基于规则推理和贝叶斯网络算法的多方证据关联分析方法,从海量司法数据中删除噪音数据,进而完成对关键信息的抽取与证据要素识别。根据数据中当事人的诉讼材料,形成结构化的证据要素,通过多方证据关联模型中证据链条实验结果与真实证据链条相似度计算结果抽取出多方证据要素的关联关系,有效实现可信证据链条的深度挖掘。At present, China’s judicial data has the problems of low data value density and poor correlation. The efficient processing of text classification from a large number of miscellaneous judicial data and the rapid extraction of effective information can not only effectively save the human and spatial resources of court litigation services, but also provide simple, safe, intelligent and efficient litigation intelligent services for the public at any time. Therefore, this paper designs a multi-party evidence association analysis method based on rule-based reasoning and Bayesian network algorithm to remove noise data from massive judicial data, and then complete the extraction of key information and the identification of evidence elements. According to the litigation materials of the parties in the data, the structured evidence elements are formed, and the correlation between the multi-evidence elements is extracted by the similarity calculation results of the evidence chain in the multi-evidence correlation model and the real evidence chain, so as to effectively realize the deep mining of the credible evidence chain.

关 键 词：司法数据 规则推理 贝叶斯网络算法 文本分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]