Object-centric process mining (OCPM) is gaining traction in both academia and industry due to its ability to model real-world processes more accurately than traditional case-centric approaches. By considering multiple interacting objects, object-centric Petri nets (OCPNs) offer a richer process representation, but this also introduces unique challenges for visualization. The presence of multiple object types, variable arcs, and complex interactions complicates the creation of clear and interpretable layouts. In this paper, we address these challenges by presenting a dedicated layout algorithm tailored to the structural characteristics of OCPNs. Inspired by the Sugiyama framework, the algorithm balances aesthetic and functional criteria, guided by a set of domain-specific quality metrics. We implemented our approach in an open-source web-based tool, OCPN Visualizer, and a reusable JavaScript library for integration into third-party applications. A user study confirms the practical relevance of our approach and highlights its effectiveness in improving the interpretability of object-centric process visualizations.
Despite the recognized importance of business process models, many processes remain unmodeled or inadequately modeled due to the time-consuming nature of manual modeling and the lack of expertise in this field. This paper proposes a novel approach for the automated process model generation, based on object analysis. It utilizes object instances, which are represented in documents and data entities, and which activities create, read, modify and consume during process execution. In contradistinction to conventional approaches that rely on event logs, our object-based approach encompasses both automated and manual activities by analyzing these tangible artifacts such as orders, invoices, and delivery notes that are already documented. In contrast, activities frequently require supplementary tracking mechanisms. The approach extracts process knowledge by identifying object types, assigning object instances to process instances, and deriving activity types from object relationships. The resulting process models are represented as high-level Petri nets, where places represent object types, transitions represent activity types, and arcs capture control and object flow while preserving object structures. Evaluation results show that this method produces more compact models than conventional discovery algorithms and identifies activities and business rules often missed by purely temporal analyses. It enables process discovery even in environments with limited system visibility or significant manual work.
The Business Process Management and the Internet of Things (BP-Meet-IoT) workshop series established itself approximately ten years ago as a platform for exploring the intersection of Business Process Management (BPM) and the Internet of Things (IoT). The workshop idea originated in a Dagstuhl seminar in 2016 and since then, the workshop has grown into a format bringing an audience to discuss a wide range of topics at the intersection of BPM and IoT. The founding of the BP-Meet-IoT workshop evolved at a time when the IoT and BPM domains themselves were mature, but its intersection remained largely unexplored. The workshop aimed to bring together the community to discuss the challenges and opportunities presented by this new paradigm. From its inception, the focus was on integrating process modeling, analysis, and automation with the specific requirements and possibilities of IoT environments. The increasing use of IoT devices, the growing complexity of interconnected systems, and the need for efficient process automation called for related research. The BPM IoT manifest [17] has become a frequently cited reference. Within the manifest that summarizes the key concepts, challenges, and future directions for combining BPM and IoT, the focus was on leveraging data from connected devices to automate, monitor, and optimize business processes.
Process mining is a powerful approach for analyzing event data, benefiting greatly from human-in-the-loop methods due to its reliance on human interpretation and decision-making. However, current tools do not fully exploit the potential of integrating process mining with interactive visual support. To address this gap, we adopt a design science research approach to systematically connect task concepts from both domains. As a result, we introduce Milana, a method that links process mining tasks, expressed as analysis questions, to established visual analytics requirements. Milana fosters a shared vocabulary, improves communication between the communities, and offers practical guidance for designing effective visualizations tailored to process mining.
Processes rely on queues to hold and prioritize cases and objects being passed between workers and teams. Consequently, analyzing queue behavior provides insights into process performance. As queue behavior is not explicitly recorded in event logs, missing information has to be restored. Existing queue mining techniques leverage queue models to restore abstractions of a queue. In this paper, we show that by treating queues and workers as objects, we can use an object-centric approach to infer missing queue information and to analyze queue behavior and performance. Evaluation on an industrial incident management process demonstrates the feasibility of the approach.
Providing accurate estimates of time-to-resolution of legal cases contributes to building trust in judicial systems. However, this task is challenging due to the temporal variability of legal proceedings and the multitude of factors influencing the time to resolution. While machine learning methods for remaining-time prediction have been extensively studied in the context of business processes, their application to judicial processes remains underexplored. A key component of remaining time prediction methods is the set of features used to train the machine learning models. In this study, we investigate the predictive power of different feature sets for estimating the time-to-resolution of legal cases in the specific context of Italian courts. We conduct an ablation study to evaluate the contribution of five categories of features: control-flow (activity sequences), temporal context, case attributes, global process state, and judge workload. The results show that, while case attributes alone perform poorly, combining them with state and temporal context significantly enhances the predictive performance. This combination achieves a mean absolute error (MAE) within an acceptable range for practical use, underscoring the value of integrating procedural and contextual information in predictive models for judicial processes.
This paper introduces VESPA (Visual Event-Stream Progressive Analytics), a framework that integrates Streaming Process Mining (SPM) with Progressive Visual Analytics (PVA) to support timely, informed decisions using partial, evolving data. VESPA addresses the challenges of analyzing transient, multifaceted event-streams by coupling process mining with progressive visualizations. Our framework is structured around key dimensions (context, task, data, algorithm, user roles, and interaction modalities). Two central research questions guide our work: identifying optimal timing for progressive visualizations and determining their effectiveness and appropriateness in streaming contexts. A prototype with ward-centric and patient-centric views was conceptualized, based on a simulated real-world scenario in the context of an emergency department (ER). These views support user roles from passive monitoring to active exploration, enabling dynamic prioritization and resource allocation. Preliminary results demonstrate the potential of VESPA to enhance situational awareness and decision-making.
Ensuring that business processes adhere to the rules imposed on them is the goal of business process compliance (BPC). The source for those rules are mainly laws, guidelines, or regulations, and consequently, BPC bridges the gap between the legal and the process domain. BPC can be structured into three phases: design-time, run-time and auditing. During each of those phases different data sources, ranging from process models to event logs, event streams and natural language texts such as process descriptions or regulatory documents, are of interest. This diversity of data sources and BPC phases requires a broad range of methods and techniques. In this keynote paper, selected approaches addressing the research vision of working towards holistic BPC are presented. Furthermore, current limitations and directions for future work are outlined, for example, how to ensure compliance of processes under changing laws and how to mitigate effects of changes to restore compliance.
The Italian lawmaking process involves a complex interplay of institutions, actors, and procedural stages. Despite the existence of previous research, systematic process-oriented studies remain scarce, largely due to the absence of structured data detailing its procedural steps. We introduce ProLiFIC (Procedural Lawmaking Flow in Italian Chambers), a machine-accessible dataset of Italian legislative preparatory works from 1985 to the present. These texts are available in unstructured form on the Normattiva portal. By leveraging large language models for event extraction, we transform this data into structured event logs suitable for analyses in process-oriented Data Science and Process Mining (PM). Our approach promotes transparency in process studies and paves the way for analyses of procedural dynamics, temporal patterns, and inefficiencies in the Italian lawmaking process. Preliminary exploratory data analyses and PM tasks demonstrate the dataset’s potential for supporting inquiries in political and legal studies.
Clinical Practice Guidelines (CPGs) are continually updated, yet translating such narrative recommendations into formal Care Pathway (CPW) processes remains labor-intensive and prone to inconsistency. We introduce a modular, end-to-end extraction pipeline that leverages Large Language Model (LLM) frameworks to automatically generate and evaluate BPMN models of CPWs from CPG text. Our system standardizes different output formats, applies LLM-based label alignment to harmonize terminology, and evaluates models against reference models using node- and structural-similarity metrics. Experiments on four stroke-related CPGs, using two state-of-the-art LLM frameworks—ProMoAI and a multi-agent orchestration approach (MAO)—shows the feasibility of automated CPW process extraction. We further observe that the multi-agent framework (MAO) demonstrates markedly higher fidelity. A web-based UI supports experiment configuration and result inspection. We released all code, prompts, and datasets as open-source to promote reproducibility and future enhancements.
In the current era of digital transformation, traditional business process management (BPM) often lack the flexibility required to manage dynamic and knowledge-intensive business processes. The purpose of this article is to present the possibilities of overcoming the limitations of traditional business process modeling focused on designing and representing the exact process flow by comparing two approaches: the widely adopted Business Process Modeling and Notation (BPMN) and the declarative, rule-based Adaptive Case Management (ACM) approach. Using a Repair Service Management (RSM) case study, we systematically analyze the strengths and limitations of both approaches regarding expressiveness, adaptability, and change management efficiency. While modeling using ad hoc subprocesses in BPMN offers some flexibility in modeling the flow of a business process, the declarative rule-based method empowers business experts to design and maintain business applications with a focus on the purpose and outcome of the business process rather than on designing its exact flow which is important to handle unforeseen situations. Building on these insights, we offer initial advice to help practitioners and researchers decide when to consider each modeling approach in dynamic business environments, contributing to the ongoing development of BPM methodologies that are essential for the digital transformation of companies.
The main aim of this paper is to analyse in a surveyed group of organisations the correlation between the level of BPM Maturity and the criteria characterising the nature of business processes. In the context of the transformation of organisations in the Industry 4.0/5.0 era, the key role of knowledge-intensive business processes is highlighted as an element supporting the innovation and adaptability of modern organisations. An empirical study was conducted in 33 organisations using two proprietary tools: the Business Process Nature Assessment Framework (BPNAF) and the BPM Maturity Assessment (BPM MA). The results indicate a lack of correlation between BPM Maturity and business process Unpredictability and Knowledge-intensity. At the same time, they indicate a positive correlation between BPM Maturity and the use of knowledge in increasing the efficiency and adaptability of semi-structured and unstructured business processes. This supports the call for integrating BPM and Knowledge Management (KM). The proposed integration promotes a better fit between BPM strategies and the realities of hyper-automation and digital duality. The article contributes to both the development of BPM theory and the practice of implementing process management and knowledge management in organisations with diverse business profiles.
Large Reasoning Models (LRMs), a subset of Large Language Models (LLMs) trained to articulate their chain-of-thought, have shown promise in tackling complex scientific tasks. However, evaluating and configuring their reasoning processes remains underexplored. This paper leverages a process mining-specific LLM evaluation framework to propose a methodology for analyzing and configuring LRMs. We introduce an approach to extract and classify reasoning steps by type (e.g., Deductive Reasoning, or Hypothesis Generation) and effect (Positive, Indifferent, Negative) on the overall reasoning, enabling a detailed assessment of reasoning quality. From this, we derive a new benchmark, PMLRM-Bench, which evaluates not only the correctness of outputs but also the robustness of the reasoning process. A case study on the QwQ-32B LLM demonstrates how targeted adjustments to reasoning type frequencies can boost task-specific performance. Our results reveal distinct reasoning patterns across models and provide actionable insights for LRM configuration. This work bridges process mining and LLM evaluation, offering a scalable framework for reasoning analysis.
Inclusivity is a key principle in Sustainable Development Goals (SDGs), especially in accelerating Digital Transformation (DT). Gender imbalance remains both understudied, particularly within the design and execution of digital projects. This study presents a Systematic Literature Review (SLR) that examines how gender imbalance factors emerges in DT projects, and how such imbalances impacts on project innovation. Based on 31 peer-reviewed studies, the review identifies three key thematic dimensions: (1) structural and institutionalized biases in digital work culture, (2) fragile support ecosystems for women in DT projects, and (3) inclusive innovation as a sustainability enabler. These themes are synthesized through the Gioia methodology and integrated into four analytical layers: women’s career decisions, organizational dynamics, DT project ecosystems, and institutional policy. The layered structure illustrates how barriers to inclusion are distributed vertically from individual level constraints to systemic policy gaps, reinforcing exclusion across the digital value chain. Findings reveal that gender equality is not only a matter of social justice but also a driver of innovation quality, team adaptability, and long term digital sustainability. This study contributes a novel perspective by positioning gender balance as a foundational design principle in digital transformation. In doing so, it offers actionable insights for researchers, practitioners, and policymakers to build more inclusive and resilient digital futures.
Organizations increasingly rely on vast amounts of unstructured textual data to describe and manage their business processes. Recent advances in Natural Language Processing (NLP) and Large Language Model (LLM)s, demonstrate significant potential for automating the extraction of process information and generating Business Process Model and Notation (BPMN) models directly from text. However, a comprehensive assessment of the methods, tools, and evaluation outcomes related to this field is missing. We present a Systematic Literature Review (SLR) that synthesizes current research on NLP-based BPMN model generation. We show which NLP approaches have been used to extract process information before and after LLMs. Analyzing 17 primary studies, we extracted information on intermediate representations of models, on supported BPMN elements, generation methods, file types, and visualizations. We also reviewed how the selected publications evaluated model generation, and derived strengths, limitations, and future directions of the field. We found that integrating rule-based or LLM feedback loops could refine model generation accuracy, and establishing benchmarks and open datasets would increase reproducibility.
Large Language Models (LLMs) are transforming how organizations manage and understand their processes. Recent conceptual studies have outlined opportunities for integrating LLMs throughout the BPM lifecycle. Prior structured reviews offer initial insights into practical application, but the rapidly evolving nature of the field requires continuous reassessment to maintain clarity on existing solutions, encountered challenges, and opportunities within each BPM lifecycle phase. Through a structured review of recent studies, we analyze how LLMs are applied across all BPM lifecycle phases. We specifically address three aspects per phase: contribution, implementation, and evaluation. Based on this, research and practical opportunities are identified. Our analysis shows that LLMs can integrate heterogeneous process documentation, generate process models from natural language, support process automation through executable outputs, and enable more accessible analysis and monitoring. Common technical approaches and recurring challenges, notably prompt sensitivity and scalability constraints, are identified. By connecting theoretical opportunities with practical applications, we offer a synthesized view of LLMs role in BPM and outline implications for future research and system development.
Inspired by recent advancements in applying Large Language Models for Business Process Management tasks, this keynote reflects on current developments in research and industry in this area. It highlights an evolution from first-wave text-extraction approaches to semantics-aware process analysis and in particular discusses the following directions: (1) moving beyond diverse text-extraction approaches that are hard to compare, by putting a growing emphasis on evaluation rigor (2) using LLMs to support data-driven process analysis by equipping them with a deeper understanding of processes, and (3) increasing the integration of LLMs as semantic connectors and coordinators for solving and supporting process analysis tasks. By focusing on these directions, the keynote aims to foster discussion on the next steps to ensure LLMs create a lasting and meaningful impact in the field of BPM.
