CAJ | 학술논문

复杂网络的社团结构对于研究现实世界有着重要意义。然而，现在的一些算法存在着划分精度不够，限制条件较多等不足，影响了复杂网络的应用。所以，本文提出了一种基于相似性复杂网络社团有效划分算法。
복잡망락적사단결구대우연구현실세계유착중요의의。연이，현재적일사산법존재착화분정도불구，한제조건교다등불족，영향료복잡망락적응용。소이，본문제출료일충기우상사성복잡망락사단유효화분산법。
The group structure of the complex network has important significance in the research of the nature. At present, however, there are some division algorithms with insufficient precision and more limited conditions etc, confining the applications of the complex network. So, this paper make use of similarity to put forward an effective group partitioning algorithm of the complex network. Applying the similarity, it can be applied with the hierarchical clustering thoughts to discover network groups by making flexible thresholds. In order to ensure the precision of the division, we define the similarity threshold at different level of the network, making the network groups preliminarily classified. Moreover, we propose the index of similarities to mine the overlapping nodes to make the structure of the network clear. As a consequence, not only the hierarchical structure of the network is found, but also the overlap nodes are discovered as well. At last, the experiments shows that the algorithm proposed in the paper enhance the accuracy of the complex group portioning.