Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Interactive Multi Interest Process Pattern Discovery

Mozhgan Vazifehdoostirani, Laura Genga, Xixi Lu, Rob Verhoeven, Hanneke van Laarhoven, Remco Dijkman

TL;DR

This work tackles the problem of discovering process patterns when analysts require multi-dimensional, outcome-oriented insights rather than a single-frequency view. It introduces the IMPresseD framework, an interactive, multi-interest PPDM that defines customizable interest functions and uses Pareto-front optimization to present non-dominated patterns, with pattern extension guided by a set of extension operators. The approach is demonstrated through a healthcare case study and multiple public logs, showing that Pareto-front–driven patterns achieve comparable or better predictive power while substantially reducing the number of patterns to inspect. The framework provides interpretable dashboards and an iterative, human-in-the-loop workflow, offering practical benefits for outcome-oriented process analysis and potential for automated deployment with reduced pattern explosion.

Abstract

Process pattern discovery methods (PPDMs) aim at identifying patterns of interest to users. Existing PPDMs typically are unsupervised and focus on a single dimension of interest, such as discovering frequent patterns. We present an interactive multi interest driven framework for process pattern discovery aimed at identifying patterns that are optimal according to a multi-dimensional analysis goal. The proposed approach is iterative and interactive, thus taking experts knowledge into account during the discovery process. The paper focuses on a concrete analysis goal, i.e., deriving process patterns that affect the process outcome. We evaluate the approach on real world event logs in both interactive and fully automated settings. The approach extracted meaningful patterns validated by expert knowledge in the interactive setting. Patterns extracted in the automated settings consistently led to prediction performance comparable to or better than patterns derived considering single interest dimensions without requiring user defined thresholds.

Interactive Multi Interest Process Pattern Discovery

TL;DR

This work tackles the problem of discovering process patterns when analysts require multi-dimensional, outcome-oriented insights rather than a single-frequency view. It introduces the IMPresseD framework, an interactive, multi-interest PPDM that defines customizable interest functions and uses Pareto-front optimization to present non-dominated patterns, with pattern extension guided by a set of extension operators. The approach is demonstrated through a healthcare case study and multiple public logs, showing that Pareto-front–driven patterns achieve comparable or better predictive power while substantially reducing the number of patterns to inspect. The framework provides interpretable dashboards and an iterative, human-in-the-loop workflow, offering practical benefits for outcome-oriented process analysis and potential for automated deployment with reduced pattern explosion.

Abstract

Process pattern discovery methods (PPDMs) aim at identifying patterns of interest to users. Existing PPDMs typically are unsupervised and focus on a single dimension of interest, such as discovering frequent patterns. We present an interactive multi interest driven framework for process pattern discovery aimed at identifying patterns that are optimal according to a multi-dimensional analysis goal. The proposed approach is iterative and interactive, thus taking experts knowledge into account during the discovery process. The paper focuses on a concrete analysis goal, i.e., deriving process patterns that affect the process outcome. We evaluate the approach on real world event logs in both interactive and fully automated settings. The approach extracted meaningful patterns validated by expert knowledge in the interactive setting. Patterns extracted in the automated settings consistently led to prediction performance comparable to or better than patterns derived considering single interest dimensions without requiring user defined thresholds.
Paper Structure (21 sections, 5 figures)

This paper contains 21 sections, 5 figures.

Table of Contents

  1. Introduction
  2. Related work
  3. Preliminaries
  4. IMPresseD framework
  5. Pattern selection
  6. Pattern extension
  7. Interest functions for outcome-oriented pattern detection
  8. Frequency interest
  9. Outcome interest
  10. Case Distance interest
  11. Implementation and evaluation
  12. User based evaluation
  13. Evaluation setup.
  14. Dataset.
  15. Results.
  16. ...and 6 more sections

Figures (5)

  • Figure 1: Overview of the IMPresseD framework
  • Figure 2: Pattern extension procedure example
  • Figure 3: Non-dominated pattern in three iterations of discovery algorithm
  • Figure 4: An example of dashboard visualization for a pattern extended from capecitabine. Note: the inner ring of pie charts and red color in distribution plots correspond to the cases with the shown pattern in the dashboard.
  • Figure 5: Quantitative evaluation results

Theorems & Definitions (5)

  • definition thmcounterdefinition: Event
  • definition thmcounterdefinition: Trace, event log
  • definition thmcounterdefinition: Partially ordered trace
  • definition thmcounterdefinition: Process pattern
  • definition thmcounterdefinition: Pattern instances set