Detekce překrývajících se komunit v sociálních sítích - perkolace klik

Abstract

This bachelor thesis focuses on algorithms for detecting overlapping communities. In detail, methods based on similar approaches such as the percolation of cliques are presented and implemented. Other methods for detecting overlapping cliques are also discussed and included in the experiments in this thesis, but are no longer implemented separately. Different modularity and conductance modifications have been selected to evaluate the quality of the detected communities. The experimental part of the work focuses on comparing the algorithms and evaluating the communities over selected datasets from social networks.