Extreme weather is often perceived as a major threat to the reliability of rail transportation. This study investigates regional rail operations in central Germany and examines whether severe weather conditions causally affect train arrival delays. We combine several years of operational data (2017–2022) with weather observations from the German Weather Service and define four component-specific treatments for adverse conditions: extreme temperature, strong wind, heavy rainfall, and snow presence. To estimate causal effects, we employ a Double Machine Learning framework based on partial linear regression (PLR–DML) and Causal Forest–DML. By reframing the weather–delay relationship as a causal question rather than a purely predictive one, the study provides evidence on whether extreme weather constitutes a materially relevant source of arrival delays. Across all four weather components, estimated average treatment effects are small in magnitude; analyses of individualized and group-level effects reveal no consistent or robust patterns of heterogeneity across lines, seasons, weekdays, or hours of the day. Robustness checks indicate that these operationally negligible estimates are not sensitive to outliers, rare-event imbalance, or to reasonable perturbations consistent with plausible unobserved confounding. Exploratory mediation analyses are consistent with the interpretation that passenger-flow variables do not materially amplify the already negligible estimated effects of severe weather on delays. Overall, the results suggest that, in this regional network, severe weather does not materially increase arrival delays. The findings underscore the importance of rigorous causal diagnostics for distinguishing perceived from materially relevant sources of delay in transportation reliability studies.
Effective resource allocation is crucial for optimizing business processes. Yet, most existing methods focus solely on single-process optimization, overlooking the interdependencies present in multi-process environments. This limitation results in inefficient resource allocation, and scalability challenges. To address this gap, we propose MuProMAC (Multi-Process Multi-Agent Coordination), a novel reinforcement learning-based method designed to optimize resource allocation across multiple interdependent business processes. Unlike prior methods, MuProMAC is the first online resource allocation method that explicitly models the interdependencies between processes and dynamically balances competing resource demands to minimize global average cycle time. We evaluate our method in five multi-process scenarios with different levels of resource contention, comparing it against state-of-the-art online resource allocation methods and three simple baselines. Our results show that MuProMAC is consistently among the top-performing methods in shared-resource environments. It achieves low cycle times and stable performance across different workload conditions, outperforming existing methods through its strong adaptability to evolving business processes and increasing complexity.
Process mining has grown into a mature research field with a wide range of techniques and applications. Much of this development builds on the pioneering work of Wil van der Aalst, whose contributions have shaped both the foundations and the growth of the discipline. Today, the field’s breadth raises the need for systematic methodological reflection to ensure that findings are robust and meaningful. In this paper, we provide a comprehensive discussion of threats to validity in process mining research. Building on the methodological framework of algorithm engineering, we analyze nine distinct validity concerns and examine how they apply across different streams of process mining. Our analysis highlights both established strengths and recurring challenges, drawing on examples from seminal contributions in the field, many inspired by Wil’s work.
Process mining aims to obtain insights from event logs through the automated analyses of recorded process data in information systems, with the ultimate aim to improve business processes running in organisations. However, real-life event logs are often incomplete, noisy, or ambiguous, such as missing timestamps or having ambiguous event labels, which traditional deterministic models cannot capture. Recent process mining developments have considered uncertainty in process mining artifacts more explicitly: in logs of recorded process behaviour, uncertainty may implicitly or explicitly influence process mining outcomes, while in process models, explicit uncertainty allows analysts to interpret and value outcomes. In this paper, we provide a conceptual foundation for uncertainty in process mining by introducing a four-level specification that separately addresses uncertainty in log attributes (e.g., activity labels of events, frequencies) and model elements (e.g., service times, read guards). For each type of uncertainty, we illustrate the levels with concrete examples to help understanding and application. We then provide a structured overview of the state of the art in stochastic process mining, classified using our specification, and present key open research challenges.
Checklists are among the most widely used instruments for guiding work, yet they have received little attention in Business Process Management (BPM) research. In 2017, we published a paper that proposed to view checklists as informational artifacts and called for a “science of checklists”. The paper proposed a conceptualization of checklists in terms of seven properties, analyzed twenty-one recurring problems with their use, and argued that a design-oriented, informational approach could be used to overcome many of these problems. The paper was published and has since been cited about thirty times. However, the research agenda it set out was not taken up and a grant proposal intended to pursue it was rejected at an early stage. In this paper, we revisit this idea as a case of an overlooked research direction. We reconstruct its argument, discuss why a published and cited idea failed to gain traction, and argue that recent developments in process mining, large language models, and AI agents make the agenda more relevant than before. Our wider aim is to ask what it means that the BPM community has paid so little attention to the artifact that arguably guides work in practice more than any other.
Process mining is a discipline that enables organizations to discover and analyze their work processes. A prerequisite for conducting a process mining initiative is the so-called event log, which is not always readily available. In such cases, extracting an event log involves various time-consuming tasks, such as creating tailor-made structured query language (SQL) scripts to extract an event log from a relational database. With this work, we investigate the use of large language models (LLMs) to support event log extraction, particularly by leveraging LLMs ability to produce SQL scripts. In this paper, we report on how effectively an LLM can assist with event log extraction for process mining. Despite the intrinsic non-deterministic nature of LLMs, our results show the potential of future LLM-assisted event log extraction tools, especially when domain and data knowledge are available. The implementation of such tools can increase access to event log extraction to a broader range of users within an organization by reducing the reliance on specialized technical skills for producing relational database query scripts and minimizing manual effort.
Research on algorithms has drastically increased in recent years. Various sub-disciplines of computer science investigate algorithms according to different objectives and standards. This plurality of the field has led to various methodological advances that have not yet been transferred to neighboring sub-disciplines. The central roadblock for a better knowledge exchange is the lack of a common methodological framework integrating the perspectives of these sub-disciplines. It is the objective of this article to develop such a research framework for algorithm engineering. Our framework builds on three areas discussed in the philosophy of science: ontology, epistemology and methodology. The framework helps us to identify and discuss various validity concerns relevant for any contribution on algorithms in various areas of computer science.
Recent advances in process mining technology have extended its applicability beyond traditional domains such as healthcare, finance and manufacturing, making it increasingly relevant for addressing problems in supply chain management. In our research, we explore the integration of process mining techniques within the domain of supply chain management, focusing on uncovering inefficiencies, ensuring compliance, and identifying opportunities for improvement. Therefore, we first review the technological advances in process mining relevant to supply chain management and outline six relevant approaches. We then combine the identified methodologies with expert interviews to derive and validate six specific use cases where process mining can significantly contribute to supply chain efficiency and resilience. The paper presents a detailed description of these use cases, demonstrating how process mining can provide actionable insights for a wide range of supply chain
problems. We discuss the implications of our findings for practitioners, who benefit from enhanced visibility and optimisation opportunities, and researchers, who are provided with a roadmap for exploration of this promising interdisciplinary field. To the best of our knowledge, this is the first work to explore potential use cases for process mining in a supply chain context, providing a comprehensive perspective on the potential benefits and challenges.
The identification of undesirable behavior in event logs is an important aspect of process mining that is often addressed by anomaly detection methods. Traditional anomaly detection methods tend to focus on statistically rare behavior and neglect the subtle difference between rarity and undesirability. The introduction of semantic anomaly detection has opened a promising avenue by identifying semantically deviant behavior. This work addresses a gap in semantic anomaly detection, which typically indicates the occurrence of an anomaly without explaining the nature of the anomaly. We propose xSemAD, an approach that uses a sequence-to-sequence model to go beyond pure identification and provides extended explanations. In essence, our approach learns constraints from a given process model repository and then checks whether these constraints hold in the considered event log. This approach not only helps understand the specifics of the undesired behavior, but also facilitates targeted corrective actions. Our experiments demonstrate that our approach outperforms existing state-of-the-art semantic anomaly detection methods.
Users generate tremendous amounts of data on the Internet every day. This so-called user-generated content (UGC) is valuable input for organizations since it may include individual experiences, opinions, and desires with respect to the products and services they offer. To automatically process UGC, automated techniques, typically referred to as Needmining, have been developed. Existing Needmining approaches extract customer needs from UGC by binarily classifying unstructured textual data into need-content and no-need content. However, they are not able to extract the specific needs. We address this research gap by developing a decision support artifact that re-conceptualizes Needmining from a binary classification problem to a token-classification problem to extract specific needs from informative content. To achieve this, we break down customer needs into components, i.e. attributes and characteristics and develop a token classification artifact. The artifact accurately identifies the need-components and, therefore, can identify specific customer needs in user-generated content. We organize and discuss the value of the artifact's output and further enrich the model with sentiment data to distinguish relevant needs. If applied, the artifact can realize efficiency gains for decisionmakers in the field of product development as it automatically and quickly identifies relevant consumer needs.
A key business challenge of process mining is to appeal to decision-makers who seek to differentiate,
with the ambition to go beyond operational optimization. One way to position process mining as a
differentiator is to integrate operational process and experience journey perspectives, with the ultimate
goal to better align operations with the needs of customers and other external stakeholders. To exemplify
this direction, this demonstration presents SAP’s journey-to-process analytics capabilities that fuse
experience with process data, allowing organizations to generate insights about how operations affect
experience.
A key business challenge of process mining is to appeal to decision-makers who seek to differentiate, with the ambition to go beyond operational optimization. One way to position process mining as a differentiator is to integrate operational process and experience journey perspectives, with the ultimate goal to better align operations with the needs of customers and other external stakeholders. To exemplify this direction, this demonstration presents SAP’s journey-to-process analytics capabilities that fuse experience with process data, allowing organizations to generate insights about how operations affect experience.
GPT-3 and several other language models (LMs) can effectively address various natural language processing (NLP) tasks, including machine translation and text summarization. Recently, they have also been successfully employed in the business process management (BPM) domain, e.g., for predictive process monitoring and process extraction from text. This, however, typically requires fine-tuning the employed LM, which, among others, necessitates large amounts of suitable training data. A possible solution to this problem is the use of prompt engineering, which leverages pre-trained LMs without fine-tuning them. Recognizing this, we argue that prompt engineering can help bring the capabilities of LMs to BPM research. We use this position paper to develop a research agenda for the use of prompt engineering for BPM research by identifying the associated potentials and challenges.
GPT-3 and several other language models (LMs) can effectively address various natural language processing (NLP) tasks, including machine translation and text summarization. Recently, they have also been successfully employed in the business process management (BPM) domain, e.g., for predictive process monitoring and process extraction from text. This, however, typically requires fine-tuning the employed LM, which, among others, necessitates large amounts of suitable training data. A possible solution to this problem is the use of prompt engineering, which leverages pre-trained LMs without fine-tuning them. Recognizing this, we argue that prompt engineering can help bring the capabilities of LMs to BPM research. We use this position paper to develop a research agenda for the use of prompt engineering for BPM research by identifying the associated potentials and challenges.
Processes in healthcare are complex and data-intensive. Process mining uses data recorded during process execution to obtain an understanding of the actual execution of a process. Due to the complexity of healthcare processes, it is useful to consider and analyse the process execution of certain cohorts, such as old and young patients, separately. While such analysis is facilitated by process variant analysis techniques, existing approaches for process variant analysis only consider a comparison based on the control flow and performance perspectives. Given the large amount of event data attributes available in healthcare settings, we propose the first data-based process variant analysis approach. Our approach allows comparing process variants based on differences in event data attributes by building on statistical tests. We applied our approach on the MIMIC-IV real-world data set on hospitalizations in the US, where we demonstrate that the approach is feasible and can actually provide relevant medical insights.
Process models play an important role in various software engineering contexts. Among others, they are used to capture business-related requirements and provide the basis for the development of process-oriented applications in low-code/no-code settings. To support modelers in creating, checking, and maintaining process models, dedicated tools are available. While these tools are generally considered as indispensable to capture process models for their later use, the initial version of a process model is often sketched on a whiteboard or a piece of paper. This has been found to have great advantages, especially with respect to communication and collaboration. It, however, also creates the need to subsequently transform the model sketch into a digital counterpart that can be further processed by modeling and analysis tools. Therefore, to automate this task, various so-called sketch recognition approaches have been defined in the past. Yet, these existing approaches are too limited for use in practice, since they, for instance, require sketches to be created on a digital device or do not address the recognition of edges or textual labels. Against this background, we use this paper to introduce Sketch2Process, the first end-to-end sketch recognition approach for process models captured using BPMN. Sketch2Process uses a neural network-based architecture to recognize the shapes, edges, and textual labels of highly expressive process models, covering 25 types of BPMN elements. To train and evaluate our approach, we created a dataset consisting of 704 hand-drawn and manually annotated BPMN models. Our experiments demonstrate that our approach is highly accurate and consistently outperforms the state of the art.
Process mining allows organizations to obtain relevant insights into the execution of their processes. However, the starting point of any process mining analysis is an event log, which is typically not readily available in practice. The extraction of event logs from the relevant databases is a manual and highly time-consuming task, and often a hurdle for the application of process mining altogether. Available support for event log extraction comes with different assumptions and requirements and only provides limited automated support. In this paper, we therefore take a novel angle at supporting event log extraction. The core idea of our paper is to use an existing process model as a starting point and automatically identify to which database tables the activities of the considered process model relate to. Based on the resulting mapping, an event log can then be extracted in an automated fashion. We use this paper to define a first approach that is able to identify such a mapping between a process model and a database. We evaluate our approach using three real-world databases and five process models from the purchase-to-pay domain. The results of our evaluation show that our approach has the potential to successfully support event log extraction based on matching.
Process automation is a crucial strategy for improving business processes, but little attention has been paid to the effects that automation has once it is operational. This paper addresses this research problem by reviewing the literature on human-automation interaction. Although many of the studies in this field have been conducted in different domains, they provide a foundation for developing propositions about process automation effects. Our analysis focuses on how humans perceive automation technology when working within a process, allowing us to propose an effective engagement model between technology, process participants, process managers, and software developers. This paper offers insights and recommendations that can help organizations optimize their use of process automation. We further derive novel research questions for a discourse within the process automation community.
Many organizations strive to increase the level of automation in their business processes. While automation historically was mainly concerned with automating physical labor, current automation efforts mostly focus on automation in a digital manner, thus targeting work that is related to the interaction between humans and computers. This type of automation, commonly referred to as business process automation, has many facets. Yet, academic literature mainly focuses on Robotic Process Automation, a specific automation capability. Recognizing that leading vendors offer automation capabilities going way beyond that, we use this paper to develop a detailed understanding of business process automation in industry. To this end, we conduct a structured market analysis of the 18 predominant vendors of business process automation solutions as identified by Gartner. As a result, we provide a comprehensive overview of the business process automation capabilities currently offered by industrial vendors. We show which types and facets of automation exist and which aspects represent promising directions for the future.
