Utilising text mining in financial fraud detection

Abstract

Financial fraud poses a significant threat to global economies with technological advancements, environmental shifts, and changes in the fraud landscape. Focusing on the intersection of text mining and fraud detection, the study aims to uncover the potential applications of text mining in financial fraud detection systems.

This thesis employs a dual approach of a systematic literature review and an inter-view-based case study. The research delves into the conceptual frameworks of big data analytics, text mining with its related fields and financial fraud detection to provide context and establish necessary understanding. The literature review introduces key research in the field alongside six case studies that focus on specific ap- plications in phishing, internal fraud, social media, loan applications and financial statement fraud.

The empirical case study examines the potential of integrating text mining into fraud detection in the case company in the banking industry. The case study finds potential in applications such as reimbursement claims.

The results indicate that text mining has great potential for fraud detection. Still, as a novel and developing field, it best serves as a complementary approach to conventional financial fraud methods. The key advantages of integrating text mining include shorter lead times for detecting fraud, improved accuracy, and the capacity to leverage more data sources. However, challenges such as cost, data privacy concerns, and limitations of textual data are notable.

The study advocates for a hybrid approach, integrating text mining with conventional fraud detection methods and other emerging technologies such as Natural Language Processing, Machine Learning and Artificial Intelligence. This strategy is proposed to address the complex dynamics of financial fraud in the modern era, aiming for a more effective and nuanced fraud detection system.