摘要： 当前的社交网络挖掘算法主要是静态的，构建社交网络的准确度不高和稳定性较差。针对上述问题，提出一种面向移动通信手机数据的动态挖掘算法。该算法在一定时间段内按照指定的时间片动态地构建社交网络，判断每一个时间片中是否存在变化点，对之前的社交网络进行合并，构建出该时间段的社交网络。对手机数据进行分析，给出满足手机数据挖掘的有向加权图。实验结果表明，与传统的Shortest Path和Random Walk算法相比，该算法可有效提高社交网络的准确度，保证社区数据增长趋势逐步达到平衡。

关键词: 社交网络, 手机数据, 动态挖掘, 时间片, 最短路径

Abstract: Aiming at the problems of the present social mining algorithms, which are primarily static and the accuracy is not high enough and stability is not enough to balance of the social networks which are constructed through present algorithms. This paper proposes a dynamic miming algorithm called Dynamic Mobile Graph Community Mining(DMGCM) which is based on mobile phone data. The main idea of this algorithm is that the social networks are dynamically constructed according to the specified time in a certain period. It is need to judge whether there is a change point in each time slice, the pervious social networks can be merged into present social networks if there exists a change point, so the complete social networks are constructed. It is necessary to analyze the mobile phone data and construct directed and weighted graph. A real dataset is used in the experiment and comparing the traditional random walk and shortest path algorithms with DMGCM algorithm. Experimental results show that algorithm effectively improves the accuracy of social networks and ensures growth trend gradually achieve the balance.

Key words: social network, mobile phone data, dynamic mining, time slice, shortest path

中图分类号: 

• TP391