Extracting Community Cores in Heterogeneous Citation Network

Abstract

The identification of closely connected cores in a complex network has become an important feature in the area of community mining in recent years. According to the network topology, it is true that a large and complex graph have a set of densely connected sub-graphs as the cores. Extraction these cores can reveal that some unexpected connection patterns between nodes in this type of network. In this paper, a simple and efficient intersection-based algorithm is proposed for finding community cores in multi-relational citation network by using the strength of modularity-based Louvain method. The proposed system is applied on a part of CiteseerX digital library [5] as the real-world data set of heterogeneous citation graph. Graph nodes represent individual papers of CiteseerX data set which are linked by three types of above relationships. The experimental results show that the proposed algorithm is highly effective in core extraction with comparable time complexity.