CAJ | 학술논문

针对传统的分裂聚类算法删除连边会影响节点划分的问题，结合K-means算法的思想并加以改进，提出了一种以邻居节点为聚类备选集，所有社团中心节点同时聚类的同步聚类算法。该算法根据节点中心度和最短路变化率确定社团中心节点集，然后以中心节点为社团中心，以邻居节点为聚类备选集合进行聚类，完成社团的划分。将算法应用到Zachary网络中并与GN算法、TopLeader算法进行比较，仿真结果表明该算法对网络有较好的划分。
침대전통적분렬취류산법산제련변회영향절점화분적문제，결합K-means산법적사상병가이개진，제출료일충이린거절점위취류비선집，소유사단중심절점동시취류적동보취류산법。해산법근거절점중심도화최단로변화솔학정사단중심절점집，연후이중심절점위사단중심，이린거절점위취류비선집합진행취류，완성사단적화분。장산법응용도Zachary망락중병여GN산법、TopLeader산법진행비교，방진결과표명해산법대망락유교호적화분。
In view of the traditional division of clustering algorithm to remove the side affects node, integrated the concept of K-means algorithm and improved it, a neighbor node as clustering optional collection, all community center node cluster synchronization clustering algorithm at the same time. The algorithm based on node degree of center and the shortest path rate determine community center node set, and then took the center node as community center, clustered the optional collection based on neighbor node, completed the division of network. Applied algorithms to Zachary network and compared with the GN algorithm and Top-Leader algorithm, the results show that this algorithm has a good division of the network.