Towards democratic group detection in complex networks.

Coscia M., Giannotti F., Pedreschi D.

Large networks  Group detection  Complex networks  Global community  Modular structures 

To detect groups in networks is an interesting problem with applications in social and security analysis. Many large networks lack a global community organization. In these cases, traditional partitioning algorithms fail to detect a hidden modular structure, assuming a global modular organization. We define a prototype for a simple localfirst approach to community discovery, namely the democratic vote of each node for the communities in its ego neighborhood. We create a preliminary test of this intuition against the state-of-the-art community discovery methods, and find that our new method outperforms them in the quality of the obtained groups, evaluated using metadata of two real world networks. We give also the intuition of the incremental nature and the limited time complexity of the proposed algorithm.

Source: Social Computing, Behavioral - Cultural Modeling and Prediction. 5th International Conference, pp. 105–113, College Park, MD, USA, 3-5 April 2012

Publisher: Springer, New York, USA

Metrics