基于Hadoop的反洗钱规则设计与实现

发布时间：2018-06-30 21:29

本文选题：反洗钱 + 大数据　；参考：《北京邮电大学》2016年硕士论文

【摘要】：随着全球经济的快速发展,金融交易越来越频繁,交易手段不断更新,交易处理的自动化、电子化的水平越来越高。现代金融产品在为广大用户提供便利的同时,也为洗钱犯罪带来了可乘之机。洗钱活动日益猖厥,引起国际社会的普遍关注,各国政府和金融监管机构也加大了对洗钱犯罪的打击力度。我国从上世纪末就开始从国家层面实施反洗钱建设,当前反洗钱监管部门建立了金融交易报告制度,为反洗钱工作的开展积累了大量数据。如何有效分析金融机构的海量交易数据和客户账户数据进而识别出洗钱交易,是反洗钱工作的关键。但是随着金融业的快速发展和金融领域信息化的不断深入,传统的反洗钱方式在处理能力和处理精度上越来越不能满足需求,同时“棱镜门”事件的爆发也暴露了 IOE等传统技术架构所隐含的安全性问题,商业银行需要一套安全、稳定、高可用的技术方案来提升自己的反洗钱能力。本文是一篇工程型论文,是以解决实际应用问题为导向的。在总结当前商业银行反洗钱业务的实际需求的基础上,对当前可用的大数据处理技术进行适用性分析,提出了一套适用于商业银行的大数据反洗钱架构,实现了基于Hadoop的反洗钱规则计算方案。总体上,本文完成了以下几个工作:(1)研究反洗钱工作在我国的进展和整体布局,总结商业银行在整个反洗钱系统中的的定位;研究商业银行的业务体系,分析商业银行反洗钱的实际需求,总结银行在反洗钱领域的不足之处;对比研究当前大数据领域的相关技术,剖析数据分析、数据存储等场景下不同大数据技术的适用情况,针对银行的实际业务需求和应用场景,设计了一套商业银行反洗钱适用的大数据处理框架。(2)研究中国人民银行关于商业银行反洗钱的相关规定,以法律法规为指导原则,结合实际的业务场景对客户行为进行分析。设计一套结合实际业务的客户行为分析指标,量化反洗钱的预警逻辑,形成一套便于计算的洗钱行为指标预警体系。(3)结合商业银行实际的业务系统和中国人民银行的监管上报要求,设计了一套反洗钱规则计算方案。将洗钱功能对业务系统进行解耦,为反洗钱的监管上报提供了数据接口,设计并完成基于Hadoop的反洗钱规则计算方案。另外,针对Hive的特性,提出了在数据模式设计、HQL代码编写、集群配置等方面的优化方案,充分释放大数据平台的数据处理能力,完成了一套高效、实用、可扩展的反洗钱规则处理系统。
[Abstract]:With the rapid development of the global economy, financial transactions become more and more frequent, the means of transaction are constantly updated, the automation of transaction processing and the level of electronization are becoming higher and higher. Modern financial products not only provide convenience for users, but also bring opportunities for money laundering crime. Money laundering is becoming more and more rampant, which has caused widespread concern in the international community. Governments and financial regulators have also stepped up the fight against money laundering. Since the end of last century, China began to carry out anti-money laundering construction at the national level. At present, the anti-money laundering supervision department has established the financial transaction reporting system, which has accumulated a lot of data for the development of anti-money laundering work. How to effectively analyze the massive transaction data and customer account data of financial institutions and identify money laundering transactions is the key to anti-money laundering work. However, with the rapid development of financial industry and the deepening of financial information, the traditional anti-money laundering method can not meet the demand in terms of processing ability and precision. At the same time, the outbreak of "Prism Gate" also exposed the security problems implied by the traditional technical framework such as IOE. Commercial banks need a set of safe, stable and high available technical solutions to enhance their anti-money laundering capability. This paper is an engineering paper, which is oriented to solve practical application problems. On the basis of summarizing the actual demand of the current commercial banks' anti-money laundering business, this paper analyzes the applicability of the current available big data processing technology, and puts forward a set of big data anti-money laundering framework suitable for commercial banks. The calculation scheme of anti-money laundering rules based on Hadoop is realized. As a whole, this paper has completed the following work: (1) to study the progress and overall layout of anti-money laundering in China, to summarize the position of commercial banks in the whole anti-money laundering system, to study the business system of commercial banks, This paper analyzes the actual demand of commercial banks in anti-money laundering, summarizes the shortcomings of banks in the field of anti-money laundering, compares and studies the relevant technologies in the current big data field, analyzes the application of different big data technologies in the scenarios of data analysis, data storage, and so on. According to the actual business needs and application scenarios of banks, this paper designs a set of big data framework for commercial banks to deal with anti-money laundering. (2) We study the relevant regulations of the people's Bank of China on anti-money laundering of commercial banks, and take the laws and regulations as the guiding principles. Combined with the actual business scenario to analyze the customer behavior. Design a set of customer behavior analysis indicators combined with actual business to quantify the early warning logic of anti-money laundering. (3) combining the actual business system of commercial banks and the supervision and reporting requirements of the people's Bank of China, a set of calculation scheme of anti-money laundering rules is designed. The money laundering function is decoupled from the business system, which provides a data interface for the supervision and reporting of anti-money laundering, and designs and completes the calculation scheme of anti-money laundering rules based on Hadoop. In addition, according to the characteristic of Hive, this paper puts forward the optimized scheme of data pattern design in HQL coding, cluster configuration and so on, which fully releases the data processing ability of big data platform, and completes a set of efficient and practical. Extensible anti-money laundering rule processing system.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F832.2;TP311.13

【参考文献】