Abstract:
Complex systems from many disciplines can be modeled by networks, specifically by nodes connected by edges. These networks exhibit a microscopic structure called "community structure". A community is seen as a subgraph composed of densely linked nodes together and weakly linked to other network nodes. The detection of this community structure is crucial to understanding the topology and operation of these networks. The majority of work in the literature concerning the community detection relate to static networks. However, many networks evolve over time. The traditional approach to community detection reuses static algorithms on different snapshots of the network and suffer of instability problems. In this work, we compare various community detection algorithms based on modularity optimization. The comparison is achieved by applying various algorithms on different datasets (increasing the number of nodes), while respecting two main aspects: time and information.
