社交网络中基于贝叶斯和半环代数模型的节点影响力计算机理

发布时间：2019-04-11 10:14

【摘要】：本文综合考虑网络结构及节点间的互动等关键因素,提出了一种节点影响力分布式计算机理.首先根据节点交互行为在时域上的自相似特性,运用带折扣因子的贝叶斯模型计算节点间的直接影响力;然后运用半环模型来分析节点间接影响力的聚合;最后根据社交网络的小世界性质及传播门限,综上计算出节点的综合影响力.仿真结果表明,本文给出的模型能有效抑制虚假粉丝导致的节点影响力波动,消除了虚假粉丝的出现对节点影响力计算带来的干扰,从中选择影响力高的若干节点作为传播源节点,可以将信息传播到更多数目的节点,促进了信息在社交网络中的传播.
[Abstract]:Considering the key factors such as network structure and interaction between nodes, a distributed computing theory of node influence is proposed in this paper. Firstly, according to the self-similarity of node interaction in time domain, the Bayesian model with discount factor is used to calculate the direct influence between nodes, and then the semi-loop model is used to analyze the aggregation of indirect influence of nodes. Finally, according to the small-world nature and propagation threshold of social networks, the comprehensive influence of nodes is calculated. The simulation results show that the proposed model can effectively suppress the fluctuation of node influence caused by false fans and eliminate the interference caused by the presence of false fans on the computation of node influence. By selecting several nodes with high influence as the source nodes, the information can be propagated to more destination nodes, which promotes the propagation of information in the social network.
【作者单位】：华中科技大学电子信息与工程系;武汉光电国家实验室;
【基金】：国家自然科学基金重点项目(批准号:61231010,60972016) 湖北省杰出青年科学家基金(批准号:2009CDA150)资助的课题~~
【分类号】：TP393.0

【参考文献】