并行最小割算法及其在金融社交网络中的应用

有效实施金融监管已成为金融健康发展的必要保证. 若能够在金融社交网络中，找到一部分承载网络中所有信息流动的关键节点，便能实现整个金融社交网络的有效监管. 金融社交网络图规模通常较大，须开发大规模图处理并行算法. 本文提出基于分布式图处理平台Pregel的并行最小割算法. 实验基于Apache Spark平台开展，所用数据均来自BoardEx数据库. 实验结果表明，在大规模社交网络图的处理中，该算法具有良好性能. 利用该并行算法得到金融社交网络图的最小割，便可有效实施金融监管.

Parallel Minimal Cut Set Algorithm and Its Application in Financial Social Networks

Effective financial supervision has become a necessary guarantee for sound development of economy. Supervising the whole financial social network effectively would become possible if a set of key nodes which carry all the information flow in the financial social network can be found. The scale of social network is often quite large, so parallel algorithms for large-scale graph processing are necessary. A Pregel-based parallel algorithm for the minimal cut set problem is proposed. The experiment is conducted on Apache Spark platform. All data used in the experiment is from the BoardEx database. Experiment results show that the algorithm has a good performance in large-scale social network graph processing. With this parallel algorithm, minimal cut sets of financial social network graphs can be obtained so that effective financial supervision can be implemented.