关键词: 社团发现, 团搜索, 二叉树, 深度遍历, [K]选择排序, MapReduce

Abstract:  For the problem of the community discovery of wireless city, during operation existing algorithm of Clique Search（CS） generates a large number of repeat groups, a redundant result and high calculation complexity and other issues. From the side storage optimization, edge treatment in advance, search generating groups, using a special binary tree storage, weights [K] selection sort, depth-first traversal algorithm to build T-CS algorithm. For massive data overflow problems, combined with MapReduce model, proposing a MP-T-CS algorithm. Experiments show that, MP-T-CS algorithm can not only solve the problem during the operation generating a large number of repeat groups, time is also greatly reduced, the cost of massive data processing capability has been greatly improved and the representativenessof groups result is greatly improved.

Key words: community discovey, group search, Binary Tree, depth traversal, [K] selection sort, MapReduce