Enhancing fraud detection in banking by integration of graph databases with machine learning.

Area of Science:

• Computer Science
• Data Science
• Financial Technology

Background:

• The banking sector's digital transformation increases fraud risks.
• Traditional fraud detection methods struggle with evolving cyber threats.
• There is a critical need for advanced fraud detection and prevention systems.

Purpose of the Study:

• To implement a state-of-the-art Graph Database approach for detecting fraudulent transactions.
• To develop and evaluate a graph-based machine learning model for real-time fraud detection.
• To enhance the security of banking operations against sophisticated fraud tactics.

Main Methods:

• Modeling relational features using Neo4j graph database.
• Applying graph-based machine learning for anomaly and pattern detection.
• Evaluating system performance using Accuracy, Recall, False Positive Rate, and ROC curves.

Main Results:

• The proposed graph-based model demonstrates effectiveness in detecting fraudulent activities.
• The approach provides immediate analysis of transactions and user behavior.
• The methodology offers a reliable framework for evaluating fraud detection systems.

Conclusions:

• Innovative technologies like graph databases are crucial for modern fraud detection.
• The study validates the effectiveness and reliability of the proposed graph-based machine learning model.
• Financial institutions can leverage this approach to strengthen their fraud prevention strategies.