基于Graphlab的网络图关键节点发现算法研究

Key nodes discovery in network graph based on Graphlab

针对桥接中心度的计算特点设计了一种分布式的网络图关键节点发现算法(DABC),并基于Graphlab进行了实现。算法具有良好的扩展性,由于能够利用集群的内存资源,算法能处理的图规模与集群的大小成正比,并且该算法利用并行处理大幅度提升了计算速度。实验表明,与传统的基于单机实现的关键节点发现算法相比,算法可以获得高达4倍的性能提升。

A distributed key nodes discovery algorithm was proposed（DABC） which was implemented on Graphlab. Due to the good scalability, the scale of graph supported by the algorithm was enlarged significantly. The parallel processing also enhances the speed of calculation. Experiment results show that proposed algorithm can achieve up to 4 times performance improvement compared with the traditional centralized key node discovery algorithm.