Investigating Similarity of Nodes' Attributes in Topological Based Communities

One of the important problems in the domain of network science is the community detection. In the past, various topological based community detection algorithms have been proposed. Recently, researchers have taken into account attributes of the nodes while proposing community detection algorithms. In this work, we investigate if the nodes in a community, identified through topology based algorithms also exhibit attribute similarity. Using four different kinds of similarity metrics, we analyse the attribute similarity of the nodes within the communities derived using five different types of topological based community detection algorithms. Based on our analysis of three real social network datasets, we found on an average of 50% attribute similarity among the nodes in the communities.