A New Multi-view and Controllable Community-uncovering Algorithm 

Huifeng LONG 

Hunan City University, Yiyang, Hunan, 413000, China 

Abstract: This paper introduces a new multi-view and controllable community-uncovering algorithm, an achievement of improving PageRank algorithm and Spin-glass model, which can avoid the overlapping com-munity structure in the process of detecting communities by means of other algorithms and also helps to im-prove the usual community-expansion model. The process of uncovering communities by the introduced algo-rithms can be divided into three steps: first, identifying the nuclear one among nodes ranked by the advanced PageRank algorithms; Second, through using multi-view recognition modularity provided by Potts spin-glass model, optimizing the expansion model of local community that is found by applying the improved Iterative Greedy algorithm to eliminate the traditional modularity’s limits in the resolution limit and the following negative effects. Finally, grasping the overlapping structure and notes carefully . By analyzing and comparing the two results of respectively using PRSGMFCA and traditional technical schemes in both computer simula-tion network and the real network , it proves that the former enjoys stronger stability and higher accuracy than the latter , and its computation complexity is also acceptable. 

Keywords: Improved PageRank algorithms, Spin-glass model, Multi-view recognition and controllable, Iterative Greedy algorithms, Community-uncovering