A Pervasive Community Detection Method Through Modular Decomposition of Markov Chain for Social Media Analysis

Abstract

In a network structure, a subset with a dense link is called a community. Many networks in the real world are composed of loosely linked multiple communities. Individual communities are compatible with a functional unit constituting a complex system represented by the network. Therefore, if the community structure of the network is clarified, it can be seen how the entire complex system is composed of individual functional units. Clarifying the community structure is a fundamental and essential task for understanding complex systems expressed in networks. Many researchers have been studying to detect communities from networks effectively and efficiently. Meanwhile, the connection density of water pipes affects the failure rate. Therefore, the connection of water pipes is regarded as a network, it is possible to estimate the trouble spot through an application of community detection. Clarifying a community of members is to demonstrate a link between members. However, autonomously formed communities in the real world have a ubiquitous (pervasive) structure. If a community has a ubiquitous structure, it implies pervasive overlapping between communities. Though studies on the community detection method (modularity maximization and infomap method divide communities into the partition that do not overlap with each other. It follows that each member belongs to only one community. Under these circumstances, in this research, we propose a new community detection method through the modular decomposition of Markov chain method for social network activity.