基于链接分析的洗钱交易识别研究

Research on Suspicious Financial Transaction Identification Based on Link Mining

如何从海量客户信息及金融交易数据中有效识别可疑金融交易,发现可靠的洗钱线索是金融机构反洗钱工作的核心问题。本文提出了一种交易路径异常链接分析模式。该模式基于图论理论,通过图形遍历方式和金融交易权重分析策略的组合,识别不同交易主体间交易活动的内在联系,发现交易流向、来源以及用途或性质异常等交易关系异常特征。交易路径异常链接分析方法不用构造频繁项目集,不用设置最小支持度和最小置信度阈值,同时具备可视化特点,在挖掘金融交易关系网络方面具有优势,为可疑金融交易识别提供有潜在价值的线索。实验结果证明了该模式的可行性和有效性。

How to identify the suspicious financial transactions and the money laundering clues effectively from massive financial transaction data is the core issue and has direct effect on anti-money laundering. For this reason, the link mining mode of abnormal transaction path identification based on graph theory is designed. By combining graph search with analysis strategy of financial transaction weight, the internal relation of transactions between differ- ent transactor can be identified, and the exceptional transactions relating to transaction direction, source, purpose and property can be found accordingly. The link mining based on graph theory doesn＇t construct frequent itemsets, nor set minimum support and confidence threshold. Accordingly, the link mining has the visualization characteristic and unique advantages in discovering relevance in financial transaction network and providing potential valuable clues for money laundering transaction identification.