Case risk identification method based on streaming and batch big data fusion calculation

Abstract

The invention discloses a case risk identification method based on streaming and batch big data fusion calculation. Firstly, based on a bank operation risk scene, corresponding data (transaction flow,operation flow and static information) is extracted and imported into a big data platform; feature extraction is carried out based on historical risk behaviors in a risk scene, and feature logic is defined; and finally, feature engineering is performed through a streaming big data calculation method and a batch big data fusion calculation method, feature combination is performed to form a rule, and finally, quick decision is performed through a Spark+Rete technology to identify suspicious risk cases. According to the method, case risk analysis is carried out on the basis of a streaming big data calculation and batch big data fusion calculation method in a massive data scene for the first time, and risk identification can be carried out quickly, flexibly and accurately.

Description

technical field [0001] The invention belongs to the technical field of data processing, and in particular relates to a case risk identification method based on stream and batch big data fusion calculation. Background technique [0002] In recent years, my country's financial operational risk cases have been frequently and frequently occurred. Many banks have successively exposed operational risk cases involving huge amounts of money. Cases related to banks' business operations have occurred frequently, and the cases in terms of operational risk have become increasingly complex. secret. Operational risk cases occur frequently, causing huge economic losses and social impact, and bringing new challenges to the stable operation of the banking industry. [0003] With decades of business derivation in banks, tens of billions of data have accumulated, how to screen suspicious risk cases in massive data and carry out effective case prevention has become an urgent problem to be solve...

AI Technical Summary

Problems solved by technology

[0004] 1) Pure batch computing has high hardware requirements, and relational databases have obvious shortcomings in terms of reading and writing, querying, and high concurrency for massive data;

[0005] 2) When the pure batch big data calculation is performed on the characteristics of massive time-series data, the calculation of the full amount of data leads to a large consumption of computing resources, a long calculation cycle, and the results of time-series calculations cannot be reused

Images