Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AgentSimulator: An Agent-based Approach for Data-driven Business Process Simulation

Lukas Kirchdorfer, Robert Blümel, Timotheus Kampik, Han van der Aa, Heiner Stuckenschmidt

TL;DR

AgentSimulator introduces a resource-first, event-log–driven approach to business process simulation by automatically inducing a multi-agent system (MAS) that models individual resource behaviors and interaction patterns. The method comprises MAS discovery (agents, types, schedules, capabilities, and handover behaviors) and a discrete-time simulation that can operate under orchestrated or autonomous handovers to produce a new event log. Across nine public datasets, AgentSimulator achieves competitive or superior accuracy across control-flow, temporal, and congestion metrics while providing substantially faster runtimes than state-of-the-art baselines, demonstrating strong adaptability to both centralized and decentralized process settings. The work highlights the practical value of agent-level interpretability for what-if analyses and proposes directions for richer agent behaviors and architectures in future research.

Abstract

Business process simulation (BPS) is a versatile technique for estimating process performance across various scenarios. Traditionally, BPS approaches employ a control-flow-first perspective by enriching a process model with simulation parameters. Although such approaches can mimic the behavior of centrally orchestrated processes, such as those supported by workflow systems, current control-flow-first approaches cannot faithfully capture the dynamics of real-world processes that involve distinct resource behavior and decentralized decision-making. Recognizing this issue, this paper introduces AgentSimulator, a resource-first BPS approach that discovers a multi-agent system from an event log, modeling distinct resource behaviors and interaction patterns to simulate the underlying process. Our experiments show that AgentSimulator achieves state-of-the-art simulation accuracy with significantly lower computation times than existing approaches while providing high interpretability and adaptability to different types of process-execution scenarios.

AgentSimulator: An Agent-based Approach for Data-driven Business Process Simulation

TL;DR

AgentSimulator introduces a resource-first, event-log–driven approach to business process simulation by automatically inducing a multi-agent system (MAS) that models individual resource behaviors and interaction patterns. The method comprises MAS discovery (agents, types, schedules, capabilities, and handover behaviors) and a discrete-time simulation that can operate under orchestrated or autonomous handovers to produce a new event log. Across nine public datasets, AgentSimulator achieves competitive or superior accuracy across control-flow, temporal, and congestion metrics while providing substantially faster runtimes than state-of-the-art baselines, demonstrating strong adaptability to both centralized and decentralized process settings. The work highlights the practical value of agent-level interpretability for what-if analyses and proposes directions for richer agent behaviors and architectures in future research.

Abstract

Business process simulation (BPS) is a versatile technique for estimating process performance across various scenarios. Traditionally, BPS approaches employ a control-flow-first perspective by enriching a process model with simulation parameters. Although such approaches can mimic the behavior of centrally orchestrated processes, such as those supported by workflow systems, current control-flow-first approaches cannot faithfully capture the dynamics of real-world processes that involve distinct resource behavior and decentralized decision-making. Recognizing this issue, this paper introduces AgentSimulator, a resource-first BPS approach that discovers a multi-agent system from an event log, modeling distinct resource behaviors and interaction patterns to simulate the underlying process. Our experiments show that AgentSimulator achieves state-of-the-art simulation accuracy with significantly lower computation times than existing approaches while providing high interpretability and adaptability to different types of process-execution scenarios.
Paper Structure (11 sections, 3 figures, 2 tables, 1 algorithm)

This paper contains 11 sections, 3 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Motivation
  3. Our Approach: AgentSimulator
  4. Definitions
  5. Phase 1: MAS Discovery
  6. Phase 2: Simulation
  7. Experiments and Results
  8. Experimental Setup
  9. Results
  10. Related Work
  11. Conclusion

Figures (3)

  • Figure 1: Sketch of the credit application process.
  • Figure 2: Overview of the end-to-end AgentSimulator approach.
  • Figure 3: Resource interactions for the BPI12W log.