Seite - 6 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

As argued in [1], applying methods originally taken from the area of business process mining [2] to human habits may represent a compromise between specification- based and learning-based methods, provided that the gap between raw sensor measurements and human actions can be filled in by performing a log preprocessing step. Such a log preprocessing step may consist of simple inferences on data or complex machine learning algorithms. 2. Applying Process Mining to Smart Spaces In business process management (BPM), a business process is a collection of related events, activities, and decisions that involve a number of actors and resources and that collectively lead to an outcome that is of value for an organization or a user. The process logic is explicitly described in terms of a process schema (i.e., the model), and a specific execution of a process is named process instance, or also case. The progress of a process instance produces a trace of execution, which may be stored in an event log and can be used for process mining [2], e.g., discovering a process model from the event log or checking the compliance of the log with the model. In order to apply a process discovery technique to a smart space, the sensor log must be turned into an event log, where the granularity chosen for the aggregation should be the same one of tasks in the process model. Additionally, the log must be segmented into traces, i.e., repetitions of the same process schema (corresponding to process cases). A basic question for the application of process mining techniques to human habits is whether human behavior is structured enough to be described using a process model. If the answer to this question is positive, the obtained model will probably resemble a “spaghetti” process, i.e., a process where the number of connections between tasks make it impossible for the model to be useful for analysis or enactment. Different approaches do exist to deal with spaghetti processes. A typical approach to deal with unstructured processes is fuzzy mining [3]; it borrows concepts from maps and cartography and apply them to zoom in and out on a process model highlighting the importance of certain tasks and connection between tasks, just like they were points and paths on a map. In this tutorial, we will (i) introduce basic concepts of process mining and fuzzy mining, (ii) how to turn a sensor log produced by a smart space into an event (action) log suitable for process mining following the methodology introduced in [4,5], and (iii) how to read the output of fuzzy mining applied to the obtained event log [6]. References [1] F. Leotta, M. Mecella, and J. Mendling, “Applying process mining to smart spaces: Perspectives and research challenges,” in Advanced Information Systems Eng. Workshops. Springer, 2015, pp. 298–304. [2] W. M. van der Aalst, Process mining data science in action. Springer, 2016. [3] C. W. Gunther and W. M. van der Aalst, “Fuzzy mining–adaptive process simplification based on multi- perspective metrics,” in Business Process Management. Springer, 2007, pp. 328–343. [4] M. Dimaggio, F. Leotta, M. Mecella, and D. Sora, “Process-based habit mining: Experiments and techniques,” in Ubiquitous Intelligence & Computing (UIC), 2016 Intl IEEE Conf., 2016, pp. 145–152. [5] F. Leotta, M. Mecella, D. Sora, and G. Spinelli, Pipelining user trajectory analysis and visual process maps for habit mining. In Ubiquitous Intelligence & Computing (UIC) 2017 Intl IEEE Conf., 2017, pp. 1-8. [6] F. Leotta, M. Mecella, and D. Sora, Visual process maps: a visualization tool for discovering habits in smart homes. Journal of Ambient Intelligence and Humanized Computing. 2019:1-29. F.LeottaandM.Mecella /Hands-onProcessMining forSmartEnvironments6