Optimization Algorithms And Its Application For Functional Module Partition

In recent years,modular structures or community detection has attracted more attention of scholars from different fields.In social networks,people within modular structures closely interact,while across different does not closely interact.In integrated circuits and communication systems,modular structures correspond to function-independent functional units.In a view of graph theory,modular structures can be considered as densely connected subgraphs.How to measure the interior and the exterior of a community impartially is of great importance for community detection.However,most measures suffer from a serious imbalance problem,which probably attributes to the measures' asymmetry between the interior and the exterior of a community,since they tend to overemphasize either the interior or the exterior of a community.In this paper,we develop a symmetric framework,which impartially measures both the interior and the exterior of a community each based on a same model to avoid the imbalance problem.We analyze our approach's resolution ability of community detection on some extreme graphs.In addition,we compare our approach with some state of the art methods on random networks of LFR(Lancichinetti,Fortunato and Radicchi)benchmark as well as real-world networks.The experimental results indicate that our approach outperforms the methods for community detection.Our approach also achieves perfect results on both the karate network and the dolphin network,which shows the accuracy of our framework for community detection.Based on the results in this paper,we can see that our framework provides us with a new platform for community detection as well as a deep understanding of characteristics of complex networks.