AI-Driven Big Data Analytics for Cyber-Threat Prediction and Risk Management in ERP Systems (SAP-style Enterprises)

Abstract

Enterprise Resource Planning (ERP) systems like SAP are the heart of information infrastructure in contemporary organisations: they combine finance, human resources, procurement, and operational processes into one platform, which is highly connected. Although this integration leads to greater efficiency and decision-making, ERP environments are also vulnerable to high risks of cybersecurity, especially insider abuse, rights expansion, and intricate cross-module attacks. The traditional rule-based security measures integrated with ERP systems are becoming less effective in recognising such threats, as they are static and have a high false-positive rate, with limited capabilities to modify behavioural patterns. This study proposes a big data analytics model based on AI-powered real-time prediction of cyber threats and risk management in the context of SAP-style ERP. The framework combines massive ingestion of logs in large volumes, distributed data processing, utilising Hadoop and Spark, and unsupervised machine-learning behavioural anomaly detection models. Engineered features to illustrate user behaviour are based on access logs, transaction history, configuration changes, and patterns of communication. Clustering-based peer analysis and autoencoder-based reconstruction error are used to detect anomalies without using labelled attack data. Identified abnormalities are combined to create a behaviour-based risk scoring system and displayed in a cybersecurity dashboard that can be used by security analysts and management for decision-making. The framework is tested on enterprise-type datasets retrieved using Kaggle and synthetically simulated attacks that are relevant in the context of ERP. Experimental analysis reveals significant advances over conventional rule-based security, such as increased detection rate, false-positive rates are significantly reduced by about 30%, and reduced incident detection and response. The findings confirm that artificial intelligence, big data analytics, and risk-oriented decision support are effective and scalable methods in enhancing ERP cybersecurity in the contemporary enterprise environment.