Process mining is an area in business process management (BPM) that provides analysts with data mining tools to discover and analyze processes from event data. Current research ofers a range of techniques to derive process insights from multiple perspectives; however, the visualization of output and its potential to facilitate a better analytical understanding has received little attention. Instead, process visualization has been treated as a means of communicating algorithmic results, primarily through mostly static control-flow-centric models that have remained largely unchanged over time. By adopting concepts from visual analytics, we can create process layouts with contextual and interactive features that are optimized for efectiveness, providing valuable reasoning tools and better support for process analysts. This paper outlines our research aimed at addressing these challenges, with the ultimate goal of writing a dissertation. We will discuss current and future contributions and how they relate to the BPM research.
Process visualization in process mining transforms records of events from information systems into
human-readable format (cf., [
A key challenge in process visualization has been handling process complexity, particularly in
processes that are large and highly variable, such as those in healthcare and Internet of Things. Process
mining has been driven by the assumption that processes are, to a certain extent, structured and
partially sequential, leaving this challenge unsolved. Furthermore, there is a narrow spectrum of visual
representations in process mining, with visualizations being mostly control-flow-centric (cf., [
In a doctoral thesis, we aim to design and evaluate novel visual techniques that extend current process visualizations with contextual attributes for multi-dimensional process analysis, interactive elements for enhanced analytical reasoning, and optimization techniques for improved efectiveness. More specifically, we first define process visualization as a layout problem that extends conventional process models to multivariate layout arrangements with a focus on time and space. Then, we introduce interactive elements to facilitate a more focused analysis and seamless navigation between multiple abstraction levels, thereby better handling visual complexity. Finally, we validate the visual representations in empirical studies to improve further using optimization techniques. In this way, we extend business process management (BPM) research by expanding the spectrum of visual representations, which are not limited to the control-flow perspective, while addressing the problem of visual complexity in large and heterogeneous datasets. We also provide a bridge between the BPM and the visual analytics communities.
This paper is structured as follows. Sec. 2 summarizes the state of the art and provide the problem statement for the thesis. Sec. 3 outlines the research design. Sec. 4 summarizes the current progress and provides an outlook.
We highlight contributions from three research streams related to this research: multi-dimensional process analysis, visual analytics and process mining, and process layout optimization.
First, process mining has primarily focused on studying control flow, but diferent sub-areas have
explored multi-perspective solutions. Process enhancement [
Second, visual analytics is a field that focuses on interactive visualizations as a means of analytical
reasoning [
Third, an efective visualization can increase cognitive comprehension [
Based on the discussion of the state of the art, we identify open challenges in process visualization that are the starting point for this thesis. First, conventional process layouts are limited to analyzing control flow, following a narrow range of analytical tasks that are visually supported. In contrast, most visualizations that take this limitation into account consider only the instance representation or deviate from model semantics (cf., [18]).
Second, process mining is based on the implicit assumption that processes are well-structured and
centralized. As a consequence, many processes, such as those from the Internet of Things and healthcare,
are not supported by standard process visuals, resulting in visually complex and cluttered outcomes.
The known "spaghetti model" is an example of such results [
Finally, there is considerable empirical evidence and expertise in process comprehension. However, only a few works explicitly aim for efective visualization based on this evidence. Even more so that evaluate for further improvement.
All in all, this thesis addresses these three challenges. We extend previous literature with novel visual techniques that define contextual, interactive, and optimized layouts in process mining.
This research envisions the development of novel visual techniques that address the limitations of conventional process models identified above by drawing on techniques from visual analytics. To this end, we build on guidelines of algorithm engineering [29, 30] and define three main objectives: RO1 Design a contextual process layout that extends the conventional process model in process mining into a multi-dimensional domain to support process analysis in answering relevant questions regarding process context without deviating from process semantics.
RO2 Design an interactive process layout that enables analysts to navigate complex, multivariate event networks with ease and minimal cognitive efort, allowing them to ask tailored analytical questions.
RO3 Design an optimized process layout based on empirical studies to enhance process comprehension, improve task performance, and facilitate conceptually and technically scalable visualizations.
The objectives build and improve upon each other. Also, each subsequent objective strengthens the connection between process mining and visual analytics, beginning with the field of process discovery. In the following subsections, we further elaborate on each objective, discuss methodology and solution, and outline our current progress.
The first objective aims to establish the visual foundations of the thesis. We focus on the classical
task of process discovery [
Two dimensions that we will focus on explicitly are the temporal and spatial perspectives, as these
are highly relevant for providing context for analytical questions (cf., [31, 32, 33]):
Time The time perspective in process mining is essential for structuring event data and answering
performance-related questions (cf., [
Contribution so far: With [35, 36], we provided the first contribution to this thesis. Here, we outlined the requirements for a timeline-based process layout [35], i.e., a process model aligned along a time axis. In [36], we further defined four layout strategies that discover a timeline-based process model examplified with the directly-follows graph while handling the visual trade-ofs from logs of varying complexities.
Space Spatial information can support analysts with a geographical context, allowing them to ask questions about where a process has been executed. This perspective is rare in process mining [37], but it is particularly relevant for specific processes, such as analyzing spatially distributed processes. For complex event logs with similar activities executed in distributed locations, geographic cues could help to provide structure. We will draw on the literature of, i.a., visualization of movement data [38] and planograms [39].
The second objective aims to build upon the previous results, expanding them into a dynamic domain.
More specifically, we define visual capabilities to allow for task-specific data analysis and utilize layout
strategies [40] to handle high-volume and high-variety data, such as filtering and abstraction. The
results are design artifacts in the form of novel interactive visual representations. We validate artifact
efectiveness in terms of process comprehension [
We will focus on variant analysis as a form of goal-oriented data analysis and pattern simplification
mechanisms:
Variation Variant analysis is an analytical task in process mining that compares process variation to
understand process performance [
Contribution so far: In Rubensson et al. [
Contribution so far: In a recent conference submission (accepted for publication at the ICPM 2025), we defined an interactive visual mining approach with non-geometric semantic zooming capabilities. This work defines a coordinate system for plotting event data on unique coordinates and introduces abstraction techniques using contour diagrams to define multiple abstraction levels. With these techniques, the user can seamlessly navigate between multiple abstraction levels of a process graph, from instance-level to process-level, and levels in-between. Process discovery and rendering occur incrementally and in real-time.
The third objective aims to build upon the previous contributions by developing an optimized and integrative solution that leverages their insights. To achieve this goal, we aim to make dedicated empirical contributions through experiments to improve and finalize the existing artifacts both conceptually and technically: Conceptual To test the overall efectiveness of our designs, we will conduct empirical studies in the form of controlled experiments with a focus on internal validity. The aim is to isolate the efects of various design decisions while controlling for user expertise. For this purpose, we will draw on experiments in software engineering [46] and reuse demographic measurement scales from prior studies (e.g., [47, 48, 49, 50]). The empirical evidence will be utilized to improve our design conceptually. Technical To ensure scalable visualizations [51], we will define requirements for incremental rendering for our designs and test their performance in computer experiments considering the guidelines by [52]. The aim is to compare our rendering strategy with others to test their impact on scalability by using relevant performance factors. The empirical evidence will be used to improve our designs technically.
In conclusion, this thesis aims to present novel visual techniques that extend conventional process visualizations in process mining with contextual information, interactive capabilities, and optimized layout arrangements. In the following, we discuss some implications.
This research makes a significant contribution to BPM and process mining. Our visual techniques extend the body of knowledge in multiple areas in process sciences, including process visualization, process discovery, process analysis, and layout comprehension. We also contribute to closing the gap between process mining and visual analytics. Finally, our implementations will be open-sourced, thus providing additional value to professionals.
The research began in July 2023 with a deadline scheduled for June 2027. As noted above, we have already provided a set of contributions to the first two objectives (Sec. 3). In the following year, we will continue with a solution for spatial visualization (Sec. 3.1), and extend and finalize the contemporary contribution of the second objective on semantic zoom (Sec. 3.2). We will also begin designing the empirical studies of objective three (Sec. 3.3), so that we can finalize them by the end of 2026.
This research presents several risks and threats. First, the research problem is challenging, stretching multiple areas in process mining. There is a risk in defining a thesis with a broad scope. To mitigate this risk, we have narrowed the scope by focusing on layout arrangements and considering only time and space as the two primary contextual aspects. We have also structured the objectives to build upon one another, providing synergies and more cohesive results.
Second, visual analytics ofers valuable tools and concepts that support this research. Process mining and visual analytics are, however, both fields with distinct academic traditions and difering expectations in terms of the type of contributions. Trying to balance the criteria of both fields is a challenging task, as there is a risk of being "stuck in the middle." To date, we have primarily focused on the process mining community as our target audience for this work.
Finally, there is limited knowledge on analytical tasks in process visualization, making validation of our artifacts challenging. We will therefore need to conduct comprehensive empirical studies, which are resource- and time-intensive. Also, visualizing spatial attributes is challenging due to the limited datasets that include these dimensions. Hence, we consider using the MIMIC-IV dataset [53], as it contains spatial information, and generate new datasets with collaborators.
The research of the author is supervised by Prof. Dr. Jan Mendling and further supported by the Einstein Foundation Berlin under grant EPP-2019-524, by the German Federal Ministry of Research, Technology and Space under grant 16DII133, and by the Deutsche Forschungsgemeinschaft under grants 496119880 (VisualMine), 531115272 (ProImpact), and 414984028 (FONDA).
[15] T. Vogelgesang, S. Rinderle-Ma, H. Appelrath, A framework for interactive multidimensional process mining, in: Business Process Management Workshops - BPM 2016 International Workshops, Rio de Janeiro, Brazil, September 19, 2016, Revised Papers, 2016, pp. 23–35. doi:10.1007/978-3-319-58457-7\_2. [16] M. T. Wynn, E. Poppe, J. Xu, A. H. M. ter Hofstede, R. Brown, A. Pini, W. M. P. van der Aalst, ProcessProfiler3D: A visualisation framework for log-based process performance comparison, Decis. Support Syst. 100 (2017) 93–108. doi:10.1016/J.DSS.2017.04.004. [17] S. Esser, D. Fahland, Multi-dimensional event data in graph databases, J. Data Semant. 10 (2021) 109–141. doi:10.1007/S13740-021-00122-1. [18] A. Yeshchenko, J. Mendling, A survey of approaches for event sequence analysis and visualization,
Inf. Syst. 120 (2024) 102283. doi:10.1016/J.IS.2023.102283. [19] A. Solti, M. Rafel, M. Stochlinski, J. Mendling, SPOT: Simulation and evaluation of real time locating system data, in: R. Clarisó, H. Leopold, J. Mendling, W. M. P. van der Aalst, A. Kumar, B. T. Pentland, M. Weske (Eds.), Proceedings of the BPM Demo Track and BPM Dissertation Award co-located with 15th International Conference on Business Process Modeling (BPM 2017), Barcelona, Spain, September 13, 2017, volume 1920 of CEUR Workshop Proceedings, CEUR-WS.org, 2017, p. "". URL: https://ceur-ws.org/Vol-1920/BPM_2017_paper_179.pdf. [20] A. Alman, A. Arleo, I. Beerepoot, A. Burattin, C. Di Ciccio, M. Resinas, Tiramisù: Making sense of multi-faceted process information through time and space, J. Intell. Inf. Syst. (2024). doi:10.1007/s10844-024-00875-8. [21] A. Hoogmoed, M. Vidgof, D. Djurica, C. Rubensson, J. Mendling, Visual representation of resource analysis insights for process mining, in: H. van der Aa, D. Bork, R. Schmidt, A. Sturm (Eds.), Enterprise, Business-Process and Information Systems Modeling - 25th International Conference, BPMDS 2024, and 29th International Conference, EMMSAD 2024, Limassol, Cyprus, June 3-4, 2024, Proceedings, volume 511 of Lecture Notes in Business Information Processing, Springer, 2024, pp. 117–128. doi:10.1007/978-3-031-61007-3\_10. [22] M. Vidgof, D. Djurica, S. Bala, J. Mendling, Interactive log-delta analysis using multi-range filtering,
Softw. Syst. Model. 21 (2022) 847–868. [23] J. Mendling, H. A. Reijers, W. M. P. van der Aalst, Seven process modeling guidelines (7PMG), Inf.
Softw. Technol. 52 (2010) 127–136. doi:10.1016/J.INFSOF.2009.08.004. [24] H. A. Reijers, J. Mendling, A study into the factors that influence the understandability of business process models, IEEE Trans. Syst. Man Cybern. Part A 41 (2011) 449–462. doi:10.1109/TSMCA. 2010.2087017. [25] D. Archambault, H. C. Purchase, The "map" in the mental map: Experimental results in dynamic graph drawing, Int. J. Hum. Comput. Stud. 71 (2013) 1044–1055. doi:10.1016/J.IJHCS.2013. 08.004. [26] P. Efinger, M. Siebenhaller, M. Kaufmann, An interactive layout tool for BPMN, in: 2009 IEEE Conference on Commerce and Enterprise Computing, CEC 2009, Vienna, Austria, July 20-23, 2009, 2009, pp. 399–406. doi:10.1109/CEC.2009.36. [27] R. J. P. Mennens, R. Scheepens, M. A. Westenberg, A stable graph layout algorithm for processes,
Comput. Graph. Forum 38 (2019) 725–737. doi:10.1111/CGF.13723. [28] W. Sonke, K. Verbeek, W. Meulemans, E. Verbeek, B. Speckmann, Optimal algorithms for compact linear layouts, in: IEEE Pacific Visualization Symposium, PacificVis 2018, Kobe, Japan, April 10-13, 2018, 2018, pp. 1–10. doi:10.1109/PACIFICVIS.2018.00010. [29] P. Sanders, Algorithm engineering - an attempt at a definition, in: S. Albers, H. Alt, S. Näher (Eds.), Eficient Algorithms, Essays Dedicated to Kurt Mehlhorn on the Occasion of His 60th Birthday, volume 5760 of Lecture Notes in Computer Science, Springer, 2009, pp. 321–340. doi:10.1007/ 978-3-642-03456-5\_22. [30] J. Mendling, H. Leopold, H. Meyerhenke, B. Depaire, Methodology of algorithm engineering,
CoRR abs/2310.18979 (2023). doi:10.48550/ARXIV.2310.18979. [31] J. L. Mennis, D. J. Peuquet, L. Qian, A conceptual framework for incorporating cognitive principles into geographical database representation, Int. J. Geogr. Inf. Sci. 14 (2000) 501–520. doi:10.1080/ 136588100415710. [32] J. Ye, S. Dobson, S. McKeever, Situation identification techniques in pervasive computing: A review,
Pervasive Mob. Comput. 8 (2012) 36–66. doi:10.1016/J.PMCJ.2011.01.004. [33] T. Grisold, H. van der Aa, S. Franzoi, S. Hartl, J. Mendling, J. vom Brocke, A context framework for sense-making of process mining results, in: 6th International Conference on Process Mining, ICPM 2024, Kgs. Lyngby, Denmark, October 14-18, 2024, 2024, pp. 57–64. doi:10.1109/ICPM63005. 2024.10680665. [34] M. Dumas, M. L. Rosa, J. Mendling, H. A. Reijers, Fundamentals of Business Process Management, 2 ed., Springer, 2018. doi:10.1007/978-3-662-56509-4. [35] H. Kaur, J. Mendling, T. Kampik, C. Rubensson, Towards timeline-based layout for process mining, in: The Practice of Enterprise Modeling - 17th IFIP Working Conference, PoEM 2024, Stockholm, Sweden, December 3-5, 2024, Proceedings, 2024, pp. 192–206. doi:10.1007/ 978-3-031-77908-4\_12. [36] C. Rubensson, H. Kaur, T. Kampik, J. Mendling, Timeline-based process discovery, Inf. Syst. 133 (2025) 102568. doi:10.1016/j.is.2025.102568. [37] S. van den Elzen, M. Jans, N. Martin, F. Pieters, C. Tominski, M. Villa-Uriol, S. J. van Zelst, Towards multi-faceted visual process analytics, Inf. Syst. 133 (2025) 102560. doi:10.1016/J.IS.2025. 102560. [38] N. V. Andrienko, G. L. Andrienko, Visual analytics of movement: An overview of methods, tools and procedures, Inf. Vis. 12 (2013) 3–24. doi:10.1177/1473871612457601. [39] A. Solti, M. Rafel, G. Romagnoli, J. Mendling, Misplaced product detection using sensor data without planograms., Decis. Support Syst. 112 (2018) 76–87. doi:10.1016/J.DSS.2018.06.006. [40] F. Du, B. Shneiderman, C. Plaisant, S. Malik, A. Perer, Coping with volume and variety in temporal event sequences: Strategies for sharpening analytic focus, IEEE Trans. Vis. Comput. Graph. 23 (2017) 1636–1649. doi:10.1109/TVCG.2016.2539960. [41] A. Gemino, Y. Wand, A framework for empirical evaluation of conceptual modeling techniques,
Requir. Eng. 9 (2004) 248–260. doi:10.1007/S00766-004-0204-6. [42] M. Sedlmair, M. D. Meyer, T. Munzner, Design study methodology: Reflections from the trenches and the stacks, IEEE Trans. Vis. Comput. Graph. 18 (2012) 2431–2440. doi:10.1109/TVCG.2012. 213. [43] F. D. Davis, Perceived usefulness, perceived ease of use, and user acceptance of information technology, MIS Q. 13 (1989) 319–340. URL: http://misq.org/ perceived-usefulness-perceived-ease-of-use-and-user-acceptance-of-information-technology. html. [44] V. Venkatesh, F. D. Davis, A theoretical extension of the technology acceptance model: Four longitudinal field studies, Manag. Sci 46 (2000) 186–204. URL: http://www.jstor.org/stable/2634758. [45] J. R. Lewis, The system usability scale: Past, present, and future, Int. J. Hum. Comput. Interact. 34 (2018) 577–590. doi:10.1080/10447318.2018.1455307. [46] C. Wohlin, P. Runeson, M. Höst, M. C. Ohlsson, B. Regnell, A. Wesslén, Experimentation in Software
Engineering, Second Edition, Springer, 2024. doi:10.1007/978-3-662-69306-3. [47] T. Kummer, J. Mendling, The efect of risk representation using colors and symbols in business process models on operational risk management performance, J. Assoc. Inf. Syst. 22 (2021) 649–694. doi:10.17705/1jais.00676. [48] J. Mendling, J. Recker, H. A. Reijers, H. Leopold, An empirical review of the connection between model viewer characteristics and the comprehension of conceptual process models, Inf. Syst.
Frontiers 21 (2019) 1111–1135. doi:10.1007/S10796-017-9823-6. [49] J. B. Gassen, J. Mendling, A. Bouzeghoub, L. H. Thom, J. P. M. de Oliveira, An experiment on an ontology-based support approach for process modeling, Inf. Softw. Technol. 83 (2017) 94–115. doi:10.1016/J.INFSOF.2016.11.005. [50] K. Figl, J. Mendling, M. Strembeck, The influence of notational deficiencies on process model comprehension, J. Assoc. Inf. Syst. 14 (2013) 312–338. doi:10.17705/1JAIS.00335. [51] S. G. Eick, A. F. Karr, Visual scalability, J. Comput. Graph. Stat. 11 (2002) 22–43. URL: http: //www.jstor.org/stable/1391126. [52] T. J. Santner, B. J. Williams, W. I. Notz, The Design and Analysis of Computer Experiments, Springer series in statistics, Springer, 2003. [53] A. Johnson, L. Bulgarelli, T. Pollard, S. Horng, L. A. Celi, R. Mark, MIMIC-IV (version 1.0), 2021. doi:10.13026/s6n6-xd98.