面向司法案件的关键要素网络感知方法研究

Key structure sensing of justiciary elements is an important problem to build the national intelligent justiciary system for nearly a year. The current research on sensing approaches does not consider structures and shows the weak theoretical basis. This project proposes an approach to key element structural network discovery about justiciary cases based on the graph-compressed sensing theoretical framework. First, most of the current anomaly sensing works focus on the anomaly point sensing research. For example, one of anomaly servers is identified quickly in the computer network. This sensing approach does not consider structures, and is not applied to anomaly structure discovery problem. This project proposes a key element network sensing approach based on structures, and solves the key element structure sensing problem. Second, the current anomaly sensing approach is proposed based on the heuristic strategy. Thus the approach does not have the theoretical basis to evaluate the key element structures. Based on the compressive sensing theoretical basis, this project focuses on building the theoretical basis to sense the key sparsity structure in the element networks. Last, from the micro-perspective of element attribute in justiciary cases, this project senses anomaly elements, analyzes the anomaly element network structure, and builds a suit of approaches to key anomaly element network structure sensing with the solid theoretical basis. From the macro-perspective of the element structure, this project makes the important tasks, such as key element extraction in justiciary cases, analysis on the mechanism of anomaly network structure, and evaluation on application setting of identified elements.

司法要素关键结构感知是近一年国家智慧司法建设遇到的一个重要问题。本项目针对现有感知方法研究中结构认识不足和理论基础薄弱，提出基于图压缩感知理论框架的司法案件关键要素网络的结构分析方法。首先，现有大量异常感知工作主要开展异常点感知研究，如：计算机网络中某一服务器异常的快速发现等，这种“忽视结构”的感知方法不适用于结构异常的发现，本项目提出一种基于结构的关键要素网络感知方法，解决要素网络结构的感知问题。其次，现有的异常感知方法主要给出启发式的感知策略，缺失评估要素网络结构质量的理论基础，本项目将基于压缩感知理论建立要素网络中“稀疏”关键结构感知的理论基础。最后，本项目将从司法案件要素属性的“微观”角度，感知要素异常，分析要素网络结构异常，构建一套具有坚实理论基础的关键要素异常结构感知方法，从要素结构的“宏观”角度，支持司法案件的关键要素抽取、分析网络结构形成机理和验证要素适用场景等重要任务

司法要素关键结构感知是近一年国家智慧司法建设遇到的一个重要问题。大量要素关键结构呈现出，跨多个要素网络，以及核心“微观”属性分布趋于正常，复杂关联时空规律。针对这一新型延伸问题，本课题设立了三个研究内容。在大规模要素网络方面，我们研究了综合利用关键结构近似计算方法，从“星型”微观结构扩展的关键结构的新型感知方法。在多源（层）要素网络方面，我们研究了利用要素特征转换的异常对齐关键结构感知方法。在要素网络时空演化方面，我们开展了面向时间序列的关键要素结构演化建模研究。项目产生了大量高质量成果，包括7篇高水平论文，其中SCI一区2篇，二区1篇，CCFA类顶级会议2篇，CCFB类重要影响会议1篇。所有论文均标记了项目号。此外，项目研究成果应用还取得了外交部感谢专函1封，天津市教委优秀研究成果三等奖1项，授权专利2项和软件著作权1项。

内容获取失败，请点击重试