Community Detection in Attributed Networks Using Graph Wavelets
Community Detection in Attributed Networks Using Graph Wavelets
复制标题
DOI：
10.1109/ieeeconf56349.2022.10051935
复制
发表时间：
2022-10
期刊：
影响因子：
--
通讯作者：
M. Ortiz-Bouza;Selin Aviyente
中科院分区：
文献类型：
--
作者：
M. Ortiz-Bouza;Selin Aviyente
许多现实世界的系统可以表示为网络，其中系统中的不同实体由节点表示，它们的交互由边表示。网络研究中的一项重要任务是社区检测，其中同一社区中的节点彼此之间的连接比与网络其余部分的连接更紧密。虽然使用节点之间的连接性（即邻接矩阵）进行社区检测已经有很多工作，但许多现实世界的网络也具有每个节点的属性信息。属性图中的社区检测需要对图结构和节点属性进行联合建模，以充分利用可用数据。在本文中，我们介绍了一种基于图信号处理的方法来进行属性网络中的社区检测。该算法使用谱图小波来过滤属性，并根据不同尺度的图过滤属性构建一个新的网络。这样，在社区检测任务中就同时考虑了图的连通性信息和节点属性。所提出的方法在多个属性社交网络上进行了评估，并且在具有二进制和数字属性的网络上表现良好。
Many real-world systems can be represented as networks where the different entities in the system are presented by nodes and their interactions by edges. An important task in the study of networks is community detection, where nodes in the same community are more densely connected to each other than they are to the rest of the network. While there has been a lot of work on community detection using the connectivity between the nodes, i.e., adjacency matrix, many real-world networks also have attribute information for each node. Community detection in attributed graphs requires joint modeling of graph structures and node attributes to make full use of available data. In this paper, we introduce a graph signal processing based approach to community detection in attributed networks. The proposed algorithm uses spectral graph wavelets to filter the attributes and constructs a new network from the graph filtered attributes across different scales. In this manner, both the graph connectivity information and the node attributes are taken into account in the community detection task. The proposed method is evaluated on multiple attributed social networks and is shown to perform well on networks with both binary and numerical attributes.