为快速准确寻找社会网络中的社区结构,从节点影响力的角度出发,提出一种新的社区发现算法.算法设计过程中引入了点权,它能够衡量节点影响力的大小;首先根据”种子”节点的点权有选择地进行广度优先搜索,使点权较大的节点不断地影响点权较小的节点,进而得到”种子”节点所在的社区结构,然后再从已知的社区外任取一个新的”种子”节点,重复上述过程,就可得到整个网络的社区结构.对算法进行优化并应用到实际网络,实验结果验证了算法的可行性,与经典算法相比,该算法的准确性和计算速度都有所提高.
In order to detect community structure in social network fast and correctly, from perspective of vertex influence, an algo- rithm for detecting local community is proposed in this paper. The algorithm introduces the vertex weight which can measure the size of vertex influence. First, we selectively detects community in Breadth First Search according to the weight of" seed" vertex, and the vertexs which have more weight can constantly influence the vertexs which have less weight,then we can get the community structure which the "seed" vertex belongs to, we search a new "seed" vertex outside the known community, then repeat this process we can get the community structure of entire network. The algorithm is optimized and its application to social network, the experiment results verifies the feasibility of the algorithm. Compared to the classical algorithm, the accuracy and computational speed of the algorithm are improved.