Dissimilarity measure for community discovery in dynamic networks

The problem of tracking the evolution of communities in dynamic networks represents one of the most recent challenging issues in the community detection field. The most of community detection approaches in a static network suffer from the instability of results. In this paper, we study the evolution of communities over time using an algorithm based on a dissimilarity measure. The main advantage of this algorithm is its automatic selection of communities with high quality in a linear execution time. Experiments on two synthetic datasets and Enron real dataset show the very good performance of the proposed algorithm when compared to DYNMOGA algorithm.