The Research Of Ego-centered Academic Community Detection Algorithm

In social network,the relationship is mapping to network structure.Then by using methods of mathematical statistics and data mining to analyze the relationship structure.Social network analysis has been used in many practical problems.Especially,the goal of the community detec-tion algorithms are to extract practical community structure from the large scale social networks according to different application requirement.Then,community detection is one of the most hot topics in social network analysis.As for the research of social community detection,most of the algorithms are all from the macro-perspective which are focus on the global social network to discovery the potential commu-nities.However,in the real applications,we usually need to extract community structure from the ego-centered perspective.The ego-centered community structure reflect one's own social communi-ty attributes which provide science guidance for the applications.In the paper,we mainly discuss an ego-centered community detection algorithm.And for the massive social data mining task,we use the MapReduce distributed computing model.And we also discuss the visualization problem of large amount data.The main job of the essay is as follows:1.This paper introduced the main research and technologies of social community detection.By analyzing the achievement,related problems and challenges of the recent research.Also,we discuss the idea and methods to meet these challenges.2.We proposed an novel ego-centered community detection algorithm via factor graph model.3.We also proposed the paralleled ego-centered community algorithm by using Hadoop and MapReduce.4.We try to analysis the effectiveness and efficiency of paralleled ego-centered community algo-rithm on different datasets.Meanwhile,we use DODO toolbox which was implemented by ourselves to complete large amount social community detection algorithm and visualization the final result will talked in detail.