CAJ | 학술논문

社团发现是复杂网络领域的一个重要的研究手段。随着网络数据规模的不断增大，现有算法难以适应较大的数据规模。针对这种情况，提出一种基于 MapReduce 的二分图社团发现算法。提出的算法可以分为两个阶段，第一个阶段将一个二分图映射为一个同质加权网络。第二个阶段利用并行化的标签传播算法来检测映射后的网络中的社团结构。在人工数据集和现实数据集中进行实验，并将提出的算法与现有的算法进行对比。实验结果表明，所提出的算法能在部分人工网络以及现实数据集中取得很好的效果，并且在算法效率上，比现有算法有很大的提高。
사단발현시복잡망락영역적일개중요적연구수단。수착망락수거규모적불단증대，현유산법난이괄응교대적수거규모。침대저충정황，제출일충기우 MapReduce 적이분도사단발현산법。제출적산법가이분위량개계단，제일개계단장일개이분도영사위일개동질가권망락。제이개계단이용병행화적표첨전파산법래검측영사후적망락중적사단결구。재인공수거집화현실수거집중진행실험，병장제출적산법여현유적산법진행대비。실험결과표명，소제출적산법능재부분인공망락이급현실수거집중취득흔호적효과，병차재산법효솔상，비현유산법유흔대적제고。
Community detection is an important research means in complex networks area.However,with the growth of networks data scale,current algorithms are hard to fit rather large-scale data.In light of such case,we propose a MapReduce-based bipartite graph community detection algorithm.The proposed algorithm can be divided into two phases.The first phase is to map a bipartite graph onto a homogeneous weighted network.The second phase is to use parallel label propagation algorithm to detect the communities in the networks mapped.Experiments have been made on synthetic datasets and real-world datasets,and the proposed algorithm is compared with existing algorithms as well.Experimental result shows that,the proposed algorithm can get quite good result in some of the synthetic networks and real-world datasets,and has big improvement in algorithm efficiency than current algorithms.