A Hierarchical Method For Content-Structured Graph Clustering

Entities in social networks, in addition to having the relationship with each other, also have content. This type of networks can be modeled by the enriched graph, in which nodes could have text too. Graph clustering is one of the important attempts toward analyzing social networks. Despite these two facts, most of the existing graph clustering methods independently focused on one of the content or structural aspects. Content-Structural graph clustering algorithms simultaneously consider both the structure and the content of the graph. The main aim of this paper is to achieve well connected (structured) clusters while their nodes benefit from homogeneous attribute values (content). The proposed algorithm in this paper so-called RSL-Cluster performs the clustering by hierarchically removing the edge between nodes which has a weight lower that the average similarity of nodes. This stage continues until reaching the user’s desired number of clusters. Comparing the proposed algorithm with three well-known content-structural clustering algorithms represents the proper functioning of the proposed method. The used measures to evaluate our method include structural, content and the content-structural measures.