Detecting Financial Crimes in Mobile Money Transactions: A Proposed Framework
Introduction
Mobile money transactions (MMT) have become increasingly popular, but with this growth comes the risk of financial crimes. In this article, we discuss a proposed framework for detecting suspicious MMT using various data mining approaches and the Dempster-Shafer theory.
Key Components of the Framework
Data Sources
- MNO log files: Historical transaction data from mobile network operators (MNOs)
- Employee records: Information about employees handling transactions
- Historical offenders’ database: A database containing information on known financial criminals
- Real-time transaction data: Live transaction data used for real-time monitoring
Pre-processing
Data cleaning and pre-processing will be used to remove incomplete and noisy transactions, ensuring that the system only processes high-quality data.
Ontology System
An ontology module will be used to set rules for identifying suspicious transactions, customer ranking, and outlier detection. This module will provide a framework for organizing and structuring knowledge about financial crimes.
Real-time Transaction Monitoring
The system will use data mining techniques such as:
- Clustering: Grouping similar transactions together
- Link analysis: Identifying connections between transactions and entities
- Location analysis: Analyzing the geographical location of transactions
These techniques will be used to check for suspicious transactions in real-time.
Alert Generator
When a suspicious transaction is detected, it will be sent to the alert generator for red flags and Suspicious Activity Reports (SAR).
Customer Profiling
Customer profile information will be used for Know Your Customer policies and Customer Due Diligence, ensuring that customers are properly vetted before engaging in transactions.
Data Warehouse
A data warehouse will be used to store information generated from monitoring for further analysis. This will provide a centralized repository for data and enable more informed decision-making.
Using the Dempster-Shafer Theory
The authors propose using the Dempster-Shafer theory to combine evidence and avoid unwanted results, such as separating genuine from illegal transactions.
Limitations of the Research
Some potential limitations of this research include:
- Continuous updating: The need for continuous updating of the rule base as new techniques in money laundering and cybercrimes emerge.
- False positives/negatives: The potential for false positives or false negatives in identifying suspicious transactions.
- Complexity: The complexity of implementing a system that can handle large volumes of data in real-time.