基于复杂网络的社交网络用户影响力研究

Research on the Influence of Social Network Users Based on Complex Network

基于复杂网络理论对社交网络用户影响力进行分析,可以为社会营销、舆情监测、信息检索等众多领域的研究提供支持。传统的网页排序算法虽然可以对有向社交网络的用户影响力进行分析,但仍存在缺陷且复杂度较高。本文提出了一种对无向社交网络进行用户影响力评价的方法,弱化了将有向网络视为无向网络研究而带来的误差,并可以高效地得到重要节点,适用范围更广。首先,本文采用网络节点的度中心性、介数中心性、接近中心性、聚类系数作为节点重要度评价指标,通过对计算数据归一化处理并取均值得到用户影响力排序的基准。其次,采用k-核分解法粗粒化地将重要度相似的节点进行归类,来检验排序的合理性。最后,通过仿真实验以及k-核分解、与HITS算法比较验证了此方法的科学性和正确性。

Based on complex network theory,analysis of social network users influence can provide support for the study of social marketing,public opinion monitoring,information retrieval,and many other fields.Although it is possible for users to analyze the influence of social networks users based on the traditional PageRank algorithm,but there is still insufficient and high complexity. This article puts forward a kind of evaluation system that evaluate the influence to social network usersbased on undirected network,weakening the error caused by regarding directed network as undirected network,and can quickly get the important nodes,applies more broadly. First of all,using node＇s degree centrality,betweenness centrality,closeness centrality,clustering coefficient as a network node importance index,calculated by the data normalization and take the mean final evaluation,as a baseline for evaluation of important network node. Secondly,the k-core decomposition method was used to test sorting through coarse-graining to classify the nodes that have similar important degree. Finally,this paper takes an example,and verifies this method of the scientific and correct through the k-core decomposition and comparingwith HITS.