Events Detection in Temporally Evolving Social Networks

Abstract

Social networks are the social structures that consist of nodes and edges. Nodes are the actors, persons, etc, and edges are the interactions among the nodes. These interactions change frequently over a time in the social networks, make temporally evolving communities. This change over a time and interactions in the communities cause evolution patterns. These evolution patterns are called as events of the Social Networks. In our paper, we detect the patterns of the interactions between the nodes and then detect the events. To achieve that goal we need to detect community. Community detection provides only structural change but it is not finding the network changes that happen over a time period. So, community mining is required to identify both the structural change and network change. In this paper, we introduce a new community mining approach to identify the similar communities and their events evolution. To do this task, we need to find current time frame t_i community changes with respect to community change in past time frame t_(i-1). To achieve this goal we are using DBLP citation dataset. This DBLP dataset represents the author and co-author relationship. In the DBLP citation dataset, we identified the existing communities and the way these communities evolve over a time period.