基于Louvain算法的复杂网络链路预测仿真

Link Prediction Simulation of Complex Network Based on Louvain Algorithm

针对复杂网络拓扑结构相似的链路预测结果精度较差、适应度不高的问题,提出基于Louvain算法的复杂网络链路预测方法。计算所有编码复杂网络中社团内部连边总值,运用Louvain算法划分复杂网络社团结构,采用标准化互信息衡量划分结果差异,明确复杂网络结构与特征;计算复杂网络拓扑结构的链路权重,将相似性指标作为共同邻居信息,考虑复杂网络边的聚类与扩散特征,利用链接拓扑权重实现权重加权预测,完成高精度复杂网络链路预测任务。实验对比结果表明,所提方法可获得较高水准的曲线下面积值,有效增强链路预测准确率与运算效率,可广泛应用于复杂网络链路预测现实场景中。

The link prediction method of complex network based on Louvain algorithm is studied in order to improve the accuracy and adaptability of the prediction results of similar links in complex network topology.The total value of the inner edge of the community in all coded complex networks was calculated.Louvain algorithm was introduced to divide the community structure of complex networks.Standardized mutual information was utilized to measure the difference of partition results,thus obtaining the structure and characteristics of complex networks.The link weight of complex network topology was also calculated.The similarity index was adopted as common neighbor information.According to the clustering and diffusion characteristics of complex network edges,the link topology weight was applied to de-weight the prediction weight.At last,the high-precision complex network link prediction task was completed.The experimental results show that the method has a high level of area value under the curve,which improves the link prediction accuracy and operation efficiency,and shows that the method has wider field applicability.