Integration of Big Data Analytics in Accounting Information Systems for Fraud Detection in the Banking Sector

Abstract

The rapid digital transformation of Indonesia's banking sector has generated a massive and complex volume of financial
transactions. The value of digital banking transactions reached IDR 52,545 trillion in 2022 (Bank Indonesia, 2023); however,
this growth has been accompanied by increasingly sophisticated fraud risks that are difficult to detect using conventional
Accounting Information Systems (AIS). Traditional systems that remain reactive relying on manual audits, rule-based systems,
and data sampling methods are often only able to detect fraud after losses have already occurred, creating an urgent need for
innovation through the integration of Big Data Analytics into AIS to support faster, more accurate, automated, and preventive
fraud detection.
This study aims to analyze how the integration of Big Data Analytics into Accounting Information Systems can improve the
effectiveness of fraud detection in Indonesia's banking sector, with a specific focus on real-time detection mechanisms, system
automation, transaction pattern analysis, and machine learning-based anomaly detection. The method employed is a
Systematic Literature Review (SLR), following a structured identification-screening-eligibility-inclusion protocol across
Scopus, Web of Science, ScienceDirect, SpringerLink, Emerald Insight, and Google Scholar, resulting in 42 articles published
between 2019 and 2025 that met the inclusion criteria.
The novelty of this study lies in synthesizing the technical fraud-detection literature which has largely focused on algorithmic
performance with the organizational and systems perspective of Accounting Information Systems, producing an integrated five
layer architecture and a real-time detection workflow that are contextualized specifically for Indonesia's banking sector rather
than adapted from developed-country settings. The findings indicate that Big Data Analytics integration enables AIS to
perform full population analysis across all financial transactions, in contrast to conventional systems that rely on sampling,
and that machine learning, anomaly detection, predictive analytics, and graph analytics consistently show higher accuracy and
lower false-positive rates across the reviewed literature, while supporting an early warning mechanism that can prevent fraud
before losses escalate.
This study concludes that integrating Big Data Analytics into Accounting Information Systems is an adaptive, preventive, and
strategically necessary step for Indonesia's banking sector during digital transformation. Successful implementation, however,
remains contingent on technology infrastructure, competent human resources, sound data governance, and stronger regulatory
support from OJK and Bank Indonesia. This study's main contribution is a practically grounded, incrementally implementable
roadmap that connects theoretical constructs (Fraud Triangle Theory, Agency Theory, TAM) with concrete architectural and
operational recommendations for Indonesian banks, including small and regional banks.
Keywords: Big Data Analytics, Accounting Information System, Fraud Detection, Real-Time Analytics, Machine Learning,
Anomaly Detection, Indonesian Banking.