基于随机游走的复杂网络聚类算法研究-计算机应用技术专业论文.docxVIP

安徽大学硕士学位论文 安徽大学硕士学位论文 Abstract Abstract In the real world，there are many complex networks，such as biotechnology network，science，technology network，social network and SO on．Recently,the study of complex networks has attracted researchers from many fields of computer，physics， mathematics，biology，etc．It has become one of the hottest interdisciplinary sciences． Complex networks clustering aims to uncover the real community structure of network，including the community structure of unsigned network，signed network， weighted network，directed network and SO on．The research of complex networks clustering has an important value of social and application，as it Can predict the networks’behaviors，facilitate the analysis of networks’topology and mine the potential functions of networks．In this paper,we make researches on the unsigned and signed complex networks，and propose the corresponding algorithms of detecting community structures based on random walk． Main works of this paper are as follows： (1)In the unsigned networks，the communities Can be divided into strong community,weak community and SO on,which is based on the obviousness of community structure．Most existing community detection algorithms show excellent performance in mining strong community structure．However,some algorithms are inadequate in mining weak community structure．This paper proposes a new method of detecting community structure based on the assumption that the probability of one node gets to other nodes inside the community is greater than the probability of other nodes outside the community．Firstly,we find the local maximum degree nodes in the global network，then，using local maximum degree nodes to find minimum complete sub-graph regarded as initial community．Nodes in initial communities are treated as clustering nodes，other nodes in network are non-clustering nodes．Secondly,calculate the probability of each non-clustering node belong to initial communities using of conditional probability model ba