社会网络中时空周期行为模式挖掘算法

摘要

A hierarchical bipartite graph based model and a mining algorithm were presented to obtain the potential spa-tio-temporal periodic behavior, meanwhile avoided the subset omitting problem in previous schemes. Then the location analysis algorithm was designed to achieve the nearly minimum location dominating subset, it could monitor the small portion of the locations as early as possible. Finally, the experiments results show that the algorithms can find out the pe-riodic location set as well as obtain nearly minimum location dominating subset, so as to cover the major portion of ob-jectives using those popular locations.%　　提出了一种层次二部图行为模式分析模型以及相应的挖掘算法，可获取潜在的时空周期行为模式，同时能克服以往算法的子集漏选问题。在此基础上，所设计的地点获取算法可以获取近似最小地点控制子集，尽早对少量地点进行监控。实验表明算法能全面地抽取周期地点子集，获取近似的最小地点控制子集，挖掘出常用地点以覆盖大部分周期行为个体。