公安情报中基于关键图谱的群体发现算法

发布时间：2018-04-30 21:39

  本文选题：群体发现 + 关键图谱　； 参考：《浙江大学学报(工学版)》2017年06期

【摘要】：为了在公安情报场景下将人的行为特征量化聚类,从而发现行为特征相似的人群并将其归类以提供决策支持,提出一种基于关键图谱的群体发现算法(KCD).KCD从人的行为特征入手,通过建立关键图谱并利用图聚类算法来进行群体发现.KCD首先将人与人之间的多个维度的行为特征进行量化计算,并将多维行为特征的量化值融合,形成三元组"人-人-值"的共现度集合;然后过滤噪音数据,建立基于行为特征的无向图;最后应用聚类算法SCAN从无向图中找出多个不同的群体,同时找出图的中心点和离群点,解决了公安情报场景中群体之间关键人物的挖掘问题.
[Abstract]:In order to quantify the behavior characteristics of people in public security information scene, we find people with similar characteristics and classify them to provide decision support, and propose a group discovery algorithm based on key map (KCD).KCD from the behavior characteristics of human, by establishing key linkage map and using graph clustering algorithm for group discovery.K CD first quantifies the behavior characteristics of multiple dimensions between human and human, and combines the quantized values of multi-dimensional behavior features to form a concurrence set of "human to human value" of three tuples, then filters noise data and establishes an undirected graph based on behavior characteristics. Finally, the clustering algorithm SCAN is used to find many different groups from the undirected graph. At the same time, we find out the central point and outlier of the graph, and solve the problem of mining key figures among public security intelligence scenes.

【作者单位】： 华中科技大学武汉光电国家实验室;
【基金】：国家自然科学基金青年基金项目(61502189)
【分类号】：D631;TP311.13
，

本文编号：1826372