基于局部扩充优化的重叠社群检测方法的改进

Improvement of overlapping community detection based on local expansion and optimization

生物信息学、社会网络、Web分析等方面的发展积累了大量的复杂网络数据信息,在对这些复杂网络进行社群检测时,往往会将一些节点归类于多个社群,目前已经提出了一些处理此类问题的算法(如LFK、GCE等),然而这类算法对局部扩充函数中参数α的选取过程复杂,无法一次性获取最优α,直接影响到了算法的可应用性。针对该缺点,提出了一种基于局部扩展的重叠社群检测的改进算法。该算法通过将α参数考虑进社群的成长过程中,使算法在保持原有速度与精度的情况下自适应地选取最优α。

Large amounts of complex networks data have been accumulated in the development of bioinformatics,social networks,Web analysis,etc. This paper tended to classify some of the nodes into multiple communities for some complex network community detection. Recently,scholars raised the issue of such processing algorithms（ such as LFM,GCE,etc.）. However,this kind of algorithm couldn＇t obtain the optimal α in one time and it had a complex process while selecting parameter α in the local expansion function which directly affected the applicability of the algorithm. To get rid of the shortage,this paper presented an improved method based on local expanding to detect overlapping community. In this method,it considered α in the growth process of communities,so that it could select parameter α in an adaptive waywhile maintaining the speed and accuracy.