中文摘要：

提出一种基于建模同步动力学行为的Kuramoto模型的网络社团发现算法SYN．该方法首先将网络中节点对象按照链接密度关系进行排序，每一个节点对象用一个一维坐标值表示，从而将网络数据矢量化．在聚类过程中，采用同步聚类原理对一个局部邻域内的对象实现同步，最终同步到一起的节点形成一个社团．通过不断扩大节点同步的邻域半径，可以得到不同分辨率的多种社团划分结果．结合社团模块度函数，可以自动选择最佳聚类结果．方法不依赖于任何数据分布假设，可以检测出任意数量、大小和形状的社团．在大量人工合成数据集和真实数据集上的实验结果表明其聚类准确率较高．

英文摘要：

A network community detection algorithm SYN, is proposed based on Kuramoto model which is a dynamic model of synchronization. Firstly, the vertices in a network are sorted according to the link densities between vertices. As a result, each vertex is projected to a one-dimensional value and the network is transformed to a vector data. During the clustering process, the data are synchronized within a local region and the data points synchronized together will be considered as a community. By enlarging the radius of synchronization, our method can detect the multi-resolution community structure of a network. Through the modularity function, our method can automatically select the optimal clustering result. Our method does not depend on any data distribution assumptions and it can detect communities of arbitrary number, size and shape in networks. The experimental results on a large number of real-world and synthetic networks show that our method achieves high accuracy.