社交网络中基于影响力的紧密子图发现算法

Social Influence Based Close Subgraph Discovery in Social Networks

随着社交网络规模的持续扩大,社交网络中社会关系的计算正逐渐成为数据挖掘研究的热点.如何高效获取社交网络中有价值的信息,如社交网络中适于投放广告的团体、客户关系中具有紧密关系的客户,具有重要的现实意义.对此,提出一个基于影响力的框架,来分析社交网络.首先,提出一种基于相似度的顶点间的权重度量.其次,通过顶点间的相似度来分配权重,改进PageRank算法计算每一个顶点潜在的影响力.最后,通过设置阈值,量化两点之间的影响力得分发现社交网络中的紧密子图.实验结果表明,基于影响力的紧密子图发现算法不仅在计算个人影响力和成员顶点间的共同影响力之间展现了很好的平衡,而且对于真实的社交网络也同样适用.

As the size of the social network continues to expand,a growing interest of social computing is arising in data mining. How to get the valuable information,such as the groups which make advertisers more properly and the related clusters in social networks,has important practical significance. In order to solve these problems,this paper presents an influence-based framework to analyze social network. First,introduce a similarity of the vertex measure. Second,allocate weights by the similarity of vertex to improve the PageRank algorithm to calculate the potential influence of each vertex. Last,through the threshold,quantify the influence between the two points score to find the close subgraph in the social network. Experimental results show that the influence-based close subgraph discovery algorithm not only shows a good balance between the calculation of personal influence score and the common influence score,but also for real social networks.