An Optimized Backbone-Based Process Layout Generation Method Using Integer Programming and Heuristics to Enhance User Comprehension

Process mining is a discipline focused on extracting valuable insights about business processes from event logs collected within information systems. A key component of process mining is visualizing process models with the aim to transform complex processes into clear and intuitive graphical representations. Such visualizations create transparency and can be overlaid with conformance and performance diagnostics. Recently, a stable graph layout algorithm has been introduced to enhance model readability by preserving the process backbone—the sequence of activities that best represents the event log. The backbone-based process layout offers a key advantage: even when filters are applied, the backbone remains intact. As a result, node positions change minimally compared to other layout methods, making it easier for users to locate relevant nodes and edges in the filtered view. Despite these benefits, several challenges remain that hinder process comprehension. In this paper, we identified key issues in existing layout methods, including unnecessary back edges, restricted horizontal alignment, non-orthogonal backbone structures, and greedy-based node placement. To address these limitations, we propose an enhanced process layout generation method that integrates integer programming with heuristic techniques. We evaluate our method using real-world event logs, showing notable reductions in back edges, edge crossings, and edge lengths, along with improvements in node orthogonality and layout symmetry. A qualitative user study further confirms a strong preference for our layout over the existing benchmark method, demonstrating its effectiveness in improving process model comprehension.