Control-flow anomaly detection by process mining-based feature extraction and dimensionality reduction

2025-01-01

Knowledge-Based Systems

AI Summary

Control-flow anomalies—such as skips and out-of-order activities—are prevalent in business process logs; however, existing consistency-checking–based methods exhibit poor robustness under noisy data and low-fidelity process models. To address this, we propose an end-to-end interpretable anomaly detection method: it first leverages a control-flow graph (CFG) generated via Heuristic Miner, embeds the CFG into a low-dimensional space using adaptive dimensionality reduction (t-SNE or UMAP), and then applies Isolation Forest for anomaly identification. This design jointly ensures semantic interpretability—by grounding features in process semantics—and strong noise resilience. Moreover, the framework supports real-time incremental detection. Evaluated on multi-source logs from the BPI Challenge, our method achieves a 12.6% improvement in F1-score and reduces false positive rate by 37%, significantly outperforming state-of-the-art approaches.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Problem

• Detect control-flow anomalies in business processes
• Improve conformance checking effectiveness using process mining
• Develop explainable framework for anomaly detection

Innovation

• Process mining-based feature extraction
• Alignment-based conformance checking
• Dimensionality reduction framework