论文部分内容阅读
社会影响力分析是社会网络研究的热点问题之一.针对社会影响力的研究,目前主要集中于社会网络中个体节点影响力的分析,而对社区级的社会影响力的研究却很少涉及.从一个全新的角度提出一种评估重叠社区影响力的方法.该方法主要包含两个部分:(1)评估每对节点(用户、实体)间的相互影响力,为了量化该影响力,提出一个相互评价学习模型MEL(Mutual Evaluation Learning);(2)基于该模型和PageRank算法的思想,设计了一种重叠社区影响力排序的算法CCIR(Cross-community Influence Rank).真实网络中的实验结果表明,该算法能够适应不同真实网络的场景,合理有效地反应真实社会网络中的社区影响力分布情况.
Analysis of social influence is one of the hot issues in social network research.For the study of social influence, the current focus on the analysis of the influence of individual nodes in social networks, but few studies on the community-level social influence. This paper proposes a new method to evaluate the influence of overlapping communities.This method mainly consists of two parts: (1) to evaluate the mutual influence between each pair of nodes (users and entities), and to quantify the influence, put forward a Mutual Evaluation Learning (MEL) model; (2) Based on the model and the idea of PageRank algorithm, a CCIR (Cross-community Influence Rank) , The algorithm can adapt to the scenes of different real networks and reasonably and effectively reflect the distribution of community influence in real social networks.