Intelligent Cross-Organizational Process Mining: A Survey and New Perspectives

TL;DR

This survey addresses a key gap in process mining: applying data-driven, cross-organizational analysis across distributed and privacy-sensitive data sources. It outlines Intelligent Cross-Organizational Process Mining (ICOPM), merging federated learning, clustering, AutoML, and (potentially) language-model guidance to enable robust, privacy-preserving insights across organizations. The paper catalogs foundational process mining concepts (workflow, discovery, conformance, enhancement), recent advances (data-centric and object-centric approaches, AutoML, infrastructure), and industrial resources (datasets, tools, metrics), then proposes an initial framework and research agenda for ICOPM. The work highlights challenges such as data privacy, data inconsistency, and explainability, while outlining opportunities in federated learning, AutoML, foundation models, and scalable computation to improve predictive accuracy, response times, and real-time decision-making in inter-organizational contexts.

Abstract

Process mining, as a high-level field in data mining, plays a crucial role in enhancing operational efficiency and decision-making across organizations. In this survey paper, we delve into the growing significance and ongoing trends in the field of process mining, advocating a specific viewpoint on its contents, application, and development in modern businesses and process management, particularly in cross-organizational settings. We first summarize the framework of process mining, common industrial applications, and the latest advances combined with artificial intelligence, such as workflow optimization, compliance checking, and performance analysis. Then, we propose a holistic framework for intelligent process analysis and outline initial methodologies in cross-organizational settings, highlighting both challenges and opportunities. This particular perspective aims to revolutionize process mining by leveraging artificial intelligence to offer sophisticated solutions for complex, multi-organizational data analysis. By integrating advanced machine learning techniques, we can enhance predictive capabilities, streamline processes, and facilitate real-time decision-making. Furthermore, we pinpoint avenues for future investigations within the research community, encouraging the exploration of innovative algorithms, data integration strategies, and privacy-preserving methods to fully harness the potential of process mining in diverse, interconnected business environments.

Figures (7)

• Figure 1: The Workflow of a Typical Process Mining System.
• Figure 2: Process Mining for Logistics.
• Figure 3: Process Mining for Healthcare.
• Figure 4: Process Mining for Finance.
• Figure 5: Process Mining for Procurement.