Recap: ICPM 2024

The International Conference on Process Mining (ICPM) is one of the leading forums for sharing the latest research and insights into process mining, an evolving field that intersects data science, business process management, and operational excellence. ICPM 2024 was held recently, showcasing a broad range of innovations, methodologies, and emerging trends that continue to shape the future of process mining.

While we couldn't attend in person, we kept a close eye on the event via Twitter and the official conference website ICPM 2024. Among the many fascinating contributions, our own Dr. Jan Mendling, a key member of our team, presented a paper that underscores our ongoing dedication to advancing process mining research. In this recap, we'll walk through some of the most intriguing papers, highlighting significant advancements and exploring their broader impact.

?

What is ICPM?

The International Conference on Process Mining (ICPM) is a premier event for academics, practitioners, and industry experts interested in process mining. The conference provides a platform for sharing cutting-edge research and innovations that advance the analysis, understanding, and optimization of business processes through data.

The 2024 edition of ICPM brought together a wide array of experts presenting novel approaches to process discovery, conformance checking, enhancement, and predictive analytics. The event’s importance lies in its role in setting the direction for future research and development in the field. We're also proud to note that Dr. Jan Mendling, a valued member of our team, submitted a paper on context-based sense-making for process mining, which we'll touch upon later.

?

Winner: AgentSimulator

One of the standout papers from this year’s ICPM was AgentSimulator: An Agent-based Approach for Data-driven Business Process Simulation by Lukas Kirchdorfer, Robert Blümel et al., receiving the Best Paper recognition. The authors introduced a novel approach to process simulation using agent-based modeling, a significant step forward compared to traditional techniques. Traditional process simulation often relies on historical data and predefined rules, focusing on modeling a process as a whole without considering the individual behaviors of participants.

In contrast, AgentSimulator models processes by simulating individual agents, each with unique characteristics and decision-making capabilities. This bottom-up approach makes it possible to capture emergent behaviors, which can provide more realistic predictions and allow for greater flexibility in understanding potential outcomes. For example, where a traditional model might treat all human resources as identical, agent-based models can simulate the unique actions of individuals responding to specific scenarios, including unpredictable variations in behavior.

The value of this approach lies in its potential to more accurately reflect real-world processes, particularly those involving human participants, where variability and autonomy are inherent. In practical terms, businesses can leverage agent-based models to gain more nuanced insights into how their processes might adapt under different conditions—providing a powerful tool for improving operational resilience and planning for contingencies.

?

Our Highlight: LLMs & Process Mining

In line with our growing interest in integrating AI capabilities into process mining, another paper that stood out to us was Evaluating the Ability of LLMs to Solve Semantics-Aware Process Mining Tasks by Adrian Rebmann et al. This research explores how large language models (LLMs), such as those underlying popular generative AI applications, can enhance process mining efforts. Specifically, the authors demonstrated the use of LLMs in providing contextual explanations for mined processes, bridging the gap between data-driven insights and human interpretability.

This topic resonates with our own direction, as we have been actively exploring how to incorporate both declarative and suggestive AI-driven features into our process mining solutions. We believe that leveraging LLMs can make complex processes more accessible and understandable, particularly for stakeholders without technical expertise, thereby adding significant value to organizational decision-making.

?

Personal Plug: Context-Making

Another notable paper from ICPM 2024 was A Context Framework for Sense-Making of Process Mining Results, submitted by, among others, our own Dr. Jan Mendling. This paper proposes a framework to enhance the interpretability of process mining results by providing contextual information that helps stakeholders better understand the insights generated.

We have previously summarized this paper in more detail on our blog, which you can read here. Dr. Mendling’s work aligns with our commitment to developing tools that facilitate deeper, more meaningful insights from process mining activities, ultimately enabling businesses to drive better outcomes.

?

Food for Thought

ICPM 2024 highlighted a number of emerging trends that are reshaping the process mining landscape. From agent-based simulation models to the integration of AI for contextual explanations, the conference provided a glimpse into the future of how businesses will leverage data to understand and optimize their processes. The emphasis on flexibility, interpretability, and human-centric modeling reflects a broader trend towards making process mining more applicable and valuable in real-world settings.

?

Conclusion

The advancements showcased at ICPM 2024 underscore the continuous evolution of process mining as a discipline. With new methodologies like agent-based modeling and AI-driven interpretability, the field is poised to become even more integral to how organizations analyze and improve their operations. As we continue our own research and development efforts, we remain committed to contributing to this exciting field and helping shape the future of process mining.

For those interested in staying ahead of the curve, we encourage you to explore the full range of papers and insights shared at ICPM 2024. The trends and technologies presented offer a roadmap for the next generation of process mining tools—tools that are more adaptable, intuitive, and effective at driving business success.

Best,

Lukas

original link: Noreja Blog, auch auf Deutsch!