Local community detection based on influence maximization in dynamic networks

Social network analysis (SNA) has opened up different research areas to researchers, such as Community Detection and Influence Maximization. By modeling social networks as graphs, one can detect one's communities or find the most Influential nodes for different applications. Despite extensive research in this area, existing methods have not yet fully met analysts' needs and are still being improved. Researchers have recently begun to apply certain concepts of a research area in social network analysis to improve social network analysis methods in other areas. In this article, we claimed that applying Two-phase Influence Maximization can improve some community detection methods. To prove the claim, we made some changes in one of the current and efficient local community detection methods to improve the way of finding the initial nodes with the new approach to finding the most influential nodes. The results showed a significant improvement. Another problem was applying this method to dynamic networks, which could be time consuming. To solve this problem, proposed a new technique that allows us to find the initial nodes in each snapshot in a new way without carrying time consuming calculations. The experimental results showed that the novel approach and the new method outperformed the previous ones in both static and dynamic social networks.

Automated summary

Social network analysis has opened up new research areas and researchers are constantly improving existing methods to meet analysts' needs. Applying the concept of Two-phase Influence Maximization can improve community detection methods, and a new technique has been proposed to solve the problem of time-consuming calculations in dynamic networks. Experimental results showed significant improvements in both static and dynamic social networks.