A reconstructed event-based framework for analyzing community evolution

摘要

People never stopped research on social networks. Researching the dynamic characteristics of social networks, we can find the key points of the network and predict the tendency of human social activities. The event-based framework is proposed by Asur et al., and it provides a good direction for analyzing the community evolution. The classification of events is the first and most important step in the framework. But we find that it has some small drawbacks to make the full framework less meaningful and applicability. This paper presents the concept of community attributes, and reconstructs each event based on the community attributes. We also use the reconstructed framework to analyze the C_DBLP and Catalano networks, and find the classification of reconstructed events is better than the original one.