Process Mining's Data Problem: Researchers Map 29 Challenges in Creating Event Logs

Process mining has emerged as a key technology for analyzing and improving business operations in the digital economy, but a fundamental bottleneck is receiving little attention: the quality and generation of the event logs that feed the entire pipeline.

The Data Quality Crisis#

Researchers conducting a systematic literature review have found that data quality stands out as the most urgent concern within the process mining community. Yet despite its criticality, the earliest and most foundational stage—creating reliable event logs from raw operational data—remains poorly understood and underexplored in academic literature.

What the Review Uncovered#

The authors examined how data-related obstacles, specifications, and existing fixes appear across event log generation work. Their findings reveal 29 distinct challenges that can be organized into 5 core requirement categories. These challenges span practical obstacles like incomplete data, inconsistent formatting, and missing timestamps, as well as deeper issues around extracting the right events from complex systems and validating that logs accurately represent actual business processes.

Engineering Approaches to Log Generation#

Current solutions already incorporate elements from two complementary domains: data engineering (infrastructure and pipeline approaches) and domain knowledge engineering (incorporating subject-matter expertise into the process). However, many practitioners lack awareness of how these fields can be systematically applied to strengthen log generation.

Why It Matters#

Flawed event logs lead to unreliable process mining outcomes, undermining improvements to business workflows and the digital systems that depend on them. By clarifying the landscape of challenges and connecting them to established engineering practices from requirements and data disciplines, the review offers researchers and practitioners a roadmap for elevating log generation quality and, in turn, the reliability of downstream process analysis.