Detekce a evoluce komunit v komplexních sítích

Abstract

This master’s thesis focuses on the implementation of algorithms for local community detection and the subsequent description of their evolution process in time frames of dynamic complex networks. The thesis works with the synthetic dynamic network that results from the optimizing SOMA algorithm and with co-author dynamic network DBLP in the time frames from 2010 to 2015. For the local community detection, four adapted algorithms were selected. Some of these algorithms can be parameterized, thus enabling the user to partially influence the resulting community structure. For the detection of evolution events, one adapted algorithm is used, and one original is defined, with respect to data variability. Both these algorithms have the set of input parameters. The thesis summarizes the experiment results of using detection and evolution algorithms and describes their differences as well.