ProMiSE: Process Mining Support for End-users

The rapid progress of digital transformation requires organizations to find intelligent ways of exploiting the vast amounts of data stored in their information systems to maintain a competitive edge. In this scenario, organizations are increasingly resorting to process mining to discover process models from event data and drive process improvement. Process mining blends principles of data mining and machine learning to extract meaningful process-related information from event data to understand and streamline processes. Over the last decade, process mining has grown into an established discipline, gaining remarkable momentum in both academia and industry. However, despite the unquestioned success of process mining, research has witnessed a lack of guidance for conducting many process mining tasks, such as creating and understanding event logs or explaining the obtained results. Indeed, some process mining tasks, such as data familiarization and process discovery, are inherently exploratory, meaning that neither the analysis goal nor the set of actions that analysts are expected to perform is precisely stated. The analysis process is unstructured and characterized by knowledge-intensive tasks executed repeatedly, as analysts often rely on their technical skills and experience to explore data. This, in turn, challenges the development of guidance for non-expert users, including novice analysts. So far, process mining research has focused mainly on technical aspects to improve algorithms and applications, while user behavior and the needs of process analysts received limited attention. However, gaining a comprehensive understanding of the analysis process, and especially of its exploratory phases, is crucial for supporting novice analysts effectively during the analysis. Thus, the central goal of this project is to gain a comprehensive understanding of how analysts do process mining in practice, i.e., the ``process of process mining'', to develop methodological guidance and operational support to assist novice analysts during the analysis effectively. To gain insights into the process of process mining, we will combine user interaction data with process mining applications (implicit feedback) and subjective insights gathered through retrospective interviews (explicit feedback). Interaction traces will be recorded with the help of a specialized tool, called ProMiSter, during observational studies on exploratory process mining tasks, conducted with process mining experts and novice analysts. For the analysis, we will combine quantitative and qualitative state-of-the-art approaches from data mining, cluster analysis, and grounded theory. The main outcome of the analysis will be a comprehensive understanding of user behavior for exploratory process mining, including frequent patterns of effective and noneffective behavior, analysis profiles, common analysis strategies, and typical challenges. Analysis patterns, profiles, and strategies will, in turn, provide strong empirical evidence for the development of methodological guidance and operational support aimed to assist novice analysts during exploratory process mining, potentially overcoming some of the identified challenges. Building upon observed effective analysis processes, we will derive a set of process mining guidelines providing practical guidance during the analysis. Besides, building upon effective and noneffective patterns of behavior, we will integrate into ProMiSter operational support in the form of recommendations and warnings based on the kind of behavior detected. By building upon the analysis of user behavior, this project will contribute to significantly advance our understanding of how analysts conduct exploratory process mining tasks, potentially providing useful empirical evidence for the improvement of process mining approaches and applications designed to support non-expert users. Besides, the obtained insights and the developed tool and guidelines will be useful to support the training and education of novice analysts and process mining students.