基于边分类的SVM模型在社区发现中的研究

摘要

社区发现是复杂网络研究的重要内容,也是分析网络结构的重要途径.分析了社区发现研究中存在的问题,提出了一种基于边分类的SVM模型.通过边顶点相似度和边介数来表示边的特征,从而构造分类函数.利用LFR生成社区结构已知的人工网络,通过人工网络数据训练基于边分类的SVM模型,对分类函数的参数进行估计,利用训练模型对真实网络进行社区分类并通过标准化互信息(NMI)和整体准确度来评价分类效果.实验得到了较高的整体准确度和NMI值.实验表明基于边分类的SVM训练模型对真实网络数据的社区划分有较高的准确度,表明该方法是可行的.%The community detection is an important part of the complex network research, and it is also the important way to analyze the network structure. In this paper,the problems existing in the community detection research are ana-lyzed and a kind of SVM model based on the edge classification is proposed. Based on vertex similarity and edge be-tweenness the characteristics of the edge are represented,so the classification function is constructed. The artificial net-work of the known community structure is generated by LFR. Through artificial network data training based on edge classification of SVM model, the parameters of classification function are estimated and the real network community is simulated by using the trained model. The higher overall accuracy and NMI values are got in the experiment. Experi-ments show that the edge classification of SVM trained model have higher accuracy on real network data and the meth-od is effective.