基于联系度的主题关注网络社区发现方法研究

发布时间：2018-03-31 09:42

本文选题：主题关注网络　切入点：集对　出处：《计算机工程与应用》2017年17期

【摘要】：目前,以兴趣或主题分享等为目的的兴趣型社交网络则引领着社交网络改革的浪潮。融合社交关系和兴趣爱好关系构建一个新型社交网络模型——主题关注模型。在此模型基础上,采用集对联系度刻画顶点间相似性度量指标,该度量方法可以更好地刻画网络结构特征,提高传统局部相似性度量指标对某些顶点间相似性值的计算精度,降低全局相似性度量指标的计算复杂度。综合考虑主题影响和社交关系,将集对联系度与凝聚型聚类算法相结合,提出一种新的主题社区发现方法。在Karate网络和豆瓣数据集上进行主题社区发现,实验结果表明,考虑主题影响的划分具有更好的社区结构。
[Abstract]:At present, interest-based social networks, such as interest or topic-sharing, are leading the reform of social networks.A new social network model-thematic focus model-is constructed by combining social relationships with interest relationships.On the basis of this model, set pair connection degree is used to describe the similarity index between vertices. This method can better describe the network structure and improve the accuracy of the traditional local similarity metrics for some vertex similarity values.Reduce the computational complexity of global similarity metrics.Considering the influence of topic and social relation, a new method of topic community discovery is proposed by combining set pair connection degree with condensed clustering algorithm.The theme community was found on the Karate network and the bean valve data set. The experimental results show that the partition considering the theme effect has a better community structure.
【作者单位】：燕山大学信息科学与工程学院;华北理工大学迁安学院;河北省虚拟技术与系统集成重点实验室;
【基金】：国家自然科学基金(No.61472340) 国家青年科学基金(No.61602401) 河北省自然科学基金(No.F2016209344) 河北省青年科学基金面上项目(No.F2017209070)
【分类号】：O157.5

【相似文献】