基于数据场分析的语义重叠社区发现算法

A semantic overlapping community detection algorithm based on semantic data fields

语义社会网络是一种由信息节点及社会关系构成的新型复杂网络,而传统社会网络社区发现算法以节点邻接关系为挖掘对象,因此无法有效处理语义社会网络重叠社区发现问题.针对这一问题,提出基于语义数据场的语义重叠社区发现算法,该算法首先以LDA(latent dirichlet allocat,ion)模型为语义信息模型,利用Gibbs取样法建立节点语义信息到语义空间的量化映射;其次,利用节点间语义坐标及链接关系,建立节点的语义数据场模型;再次,以语义关系强度及语义势能为参数,提出一种改进的语义社会网络重叠社区发现的随机游走策略;最后提出可度量语义社区发现结果的语义模块度模型.通过实验分析,验证了本文算法及语义模块度模型的有效性及可行性.

Semantic social networks （SSNs） represent a new type of complex network; consequently, they cannot be analyzed efficiently by traditional community detection algorithms that depend on social network adjacency. To solve this problem, an overlapping community structure-detecting method for SSNs is proposed, and is based on semantic data fields. First, the paper proposes an algorithm that utilizes Gibbs sampling to create the quanti- zation mapping that enables semantic information in nodes to be moved into semantic space, using LDA （latent dirichlet allocation） as the semantic model. Second, it establishes a semantic data field model, using the semantic coordinates and link relationships of nodes. Third, it proposes an improved random walk strategy that employs an overlapping community structure-detecting algorithm for SSNs using the semantic relationship strength and the semantic potential of nodes as parameters. Finally, it proposes the semantic model by which an SSN community structure can be measured. The efficiency, feasibility, and semantic modularity of the algorithm is verified by experimental analysis.