A multi-agent system is a useful modeling architecture in business process modeling in the sense that we can naturally implement participants in a real company with software agents. However, analyzing and interpreting the simulation results of multi-agent models tends to be difficult due to the inherent complexity of the models. In this regard, another discipline—process mining—is useful for such purposes because it has demonstrated its usefulness in analyzing real processes. In this article, our aim is to combine these two disciplines for exploitation in business process modeling and simulation; we extend a multi-agent-based business simulator named Multi-Agent system with Resource-Event-Agent ontology (MAREA) to be able to be analyzed by means of process mining techniques. To this end, we formalize the abstract multi-agent architecture of MAREA and establish its relationship to process mining by defining how execution of a multi-agent system can be recorded as an event log, which is later analyzed by process mining techniques. Based on this definition, we implement functionality to extract event logs from simulation runs in MAREA. For demonstration, we implement a model of a generic trading company in MAREA and perform process structure verification and social network analyzes by means of process mining techniques.

中文翻译：

---

多代理业务模拟器的流程挖掘

在我们可以自然地利用软件代理在真实公司中实现参与者的意义上，多代理系统是业务流程建模中有用的建模架构。然而，由于模型固有的复杂性，分析和解释多主体模型的仿真结果往往很困难。在这方面，另一个学科（过程挖掘）对于此类目的很有用，因为它已经证明了其在分析实际过程中的有用性。在本文中，我们的目标是将这两个学科结合起来以进行业务流程建模和仿真的开发。我们扩展了具有资源-事件-代理本体（MAREA）的基于多代理的业务模拟器Multi-Agent系统，使其能够通过流程挖掘技术进行分析。为此，我们将MAREA的抽象多主体体系结构形式化，并通过定义如何将多主体系统的执行记录为事件日志来建立其与流程挖掘的关系，然后通过流程挖掘技术对其进行分析。基于此定义，我们实现了从MAREA中的模拟运行中提取事件日志的功能。为了演示，我们在MAREA中实现了一家通用贸易公司的模型，并通过过程挖掘技术进行了过程结构验证和社交网络分析。