Micro data model architecture for AML scoring rule engines

Abstract

Online and mobile banking have become a primary service of today’s banking and financial sector. Clients could do their primary transactional jobs without physically appearing on the bank. This facility is 24x7 available. So, detection of money laundering activities based on transactional data analysis is a key challengeable area in today’s banking and financial sector. Businesses are trying to prevent money laundering activities by applying rule-based techniques to the real time operational transactions which could not completely cure the problem because higher constraints on the operational transaction could inconvenience the legal customer base and lose the customer satisfaction over the time. So, the near-real time and traditional data warehousing approaches with post detection techniques becomes the most common approach to detect money laundering activities in today’s banking and financial context. Traditional data warehousing approaches loaded data from operational or transactional systems on a weekly or nightly basis. Near real-time and real-time data warehouse approaches use real-time ETL tools to load data into the data warehouse in predefined shorter time intervals which preserve a gap with real-time transactional data. In addition to that, running anomaly detection engines (rule based or machine learning models) on top of those massive amounts of data (either OLTP databases or warehouse database) will take another considerable time due to higher velocity of data. So, identifying money launderers by analyzing post detection techniques causes higher risk to the financial system because the money launderer may leave the financial system before the money launderer catches. This report introduce a novel data modelling architecture named “Micro Data Model Architecture” and an associated supporting tool named “Micro Temporal Database Generator” for “scoring rule engines” to detect financial fraudulent activities earlier by removing the burden on operational data sources.