基于链路预测的小微授信客户隐性关联关系识别研究  

Research on Identification of Implicit Correlations of Small and Micro Enterprises Based on Link Prediction

作　　者：程广华[1] CHENG Guanghua(Bank of Communications,Shanghai 200120,China)

机构地区：[1]交通银行总行,上海200120

出　　处：《江苏工程职业技术学院学报》2018年第3期5-9,共5页Journal of Jiangsu College of Engineering and Technology

摘　　要：商业银行在小微业务信贷风险管理实践中的核心问题是:如何识别小微企业客户间的隐性关联关系,全面准确地评估风险,做好资产质量管控。从风险管理的实践看,小微企业是以人为核心的微观经济主体,企业的日常管理、资金运作等紧紧掌握在一些核心人员手中,除显性的股权、任职、上下游关系外,核心人员在银行不同渠道留存的信息也构成了复杂网络。基于此类信息,采用链路预测算法,使用Python的net-workx库,识别不同人员之间存在链接的可能性,进而识别不同人员控制的企业间的隐性关联关系。The core problem of credit risk management for commercial banks is how to identify the implicit correlations of small and micro enterprises,thereby carrying out comprehensive and accurate evaluations of risks,and making good management of asset quality. By analyzing the practice of risk management,it is clearly shown that daily management and capital operation of the micro-enterprises,which are people-centered micro-economic entities, are tightly controlled by nuclear persons whose information kept in banks through different channels,altogether with the explicit ownership,employment,and upstream and downstream relationships,forms complicated networks. Based on link prediction and Python’s networkx library,the possibility of links among different people can be ascertained,and thus the implicit correlations of enterprises controlled by them can be identified.

关 键 词：商业银行 小微企业 关联关系 复杂网络 链路预测 PYTHON 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术] F832.4[自动化与计算机技术—计算机科学与技术]