CAJ | 학술논문

提出了一种重叠社区发现的启发式算法.该算法基于局部贡献度的思想,以度最大的节点作为初始社区,逐步把对社区贡献最大的邻节点加入社区;同时考虑了社区的重叠性,若存在对多个社区贡献都很大的边界节点,则把边界节点同时加入到这些社区中.最后利用重叠系数对所划分的社区进行调整,使社区结构更加合理.对两个经典的社会网络Zachary和American CollegeFootball进行了实验测试,实验结果表明:该算法能快速准确地划分出社区,并能挖掘出社区间的边界节点.
제출료일충중첩사구발현적계발식산법.해산법기우국부공헌도적사상,이도최대적절점작위초시사구,축보파대사구공헌최대적린절점가입사구;동시고필료사구적중첩성,약존재대다개사구공헌도흔대적변계절점,칙파변계절점동시가입도저사사구중.최후이용중첩계수대소화분적사구진행조정,사사구결구경가합리.대량개경전적사회망락Zachary화American CollegeFootball진행료실험측시,실험결과표명:해산법능쾌속준학지화분출사구,병능알굴출사구간적변계절점.
A heuristic algorithm is proposed to detect overlapping communities in this paper.The algorithm is based on the local modularity and the vertex with greatest degree is considered as the initial community.Then expanding the community by putting the adjacent vertex which has maximal contribution into it.Furthermore,the algorithm takes into account the overlapping property of community.If there are border vertices which contribute greatly to more than one community,adding them to these communi-ties.Finally,In order to make the result more reasonable,the detected communities is adjusted according to overlapping coefficient. Two typical social networks,Zachary and American College Football,are applied to the algorithm.The results show that it can de-tect overlapping communities rapidly and correctly and also mine the border vertices.