论文部分内容阅读
社交网络用户影响力度量是意见领袖识别的必要前提,然而目前的度量模型忽略了多维影响因素以及水军群体对度量模型真实性的影响.为此本文从网络结构、交互行为和交互信息三个维度来分析整合相关的影响因素,基于LeaderRank模型构建多维用户影响力度量-MUI模型,并在水军识别的基础上构建信任惩罚模型,以修正MUI模型建立MUISTP模型,实现社交网络用户影响力的真实性度量.以新浪微博平台大规模实际数据进行实验分析结果表明,与FBI,LeaderRank和Generative Graphical模型相比,MUI模型识别出的意见领袖更为准确,且可以实现更为有效的信息扩散.此外,经过水军信任惩罚后MUISTP模型较MUI能够实现更为有效的意见领袖识别.
However, the current measurement model ignores the multidimensional influencing factors and the influence of the navy group on the authenticity of the measurement model.Therefore, this article from the network structure, interactive behavior and interactive information of three Dimension to analyze the influencing factors of integration, build a multi-dimensional user influence measure-MUI model based on the LeaderRank model, build a trust penalty model based on the water army identification, modify the MUI model to establish the MUISTP model, and realize the influence of the social network users Authenticity measurement.The experimental analysis on the large-scale real data of Sina Weibo platform shows that the opinion leaders identified by the MUI model are more accurate and can achieve more effective information diffusion than the FBI, LeaderRank and Generative Graphical models In addition, MUISTP model can achieve more effective opinion leader recognition than MUI after the naval trust punishment.