Page - 5 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

Hands-on Process Mining for Smart Environments Francesco Leotta and Massimo Mecella a Sapienza Università di Roma {leotta,mecella}@diag.uniroma1.it Abstract. A software system managing a smart space takes, among its inputs, models of human behavior; such models are usually difficult to obtain and to validate. The employment of techniques from business process modeling and mining may represent a solution to both the problems, but a set of challenges need to be faced in order to cope with major differences between human activities and business processes. As a first step, this tutorial introduces basic concepts of business process management (BPM) and process mining. With these concepts in mind, the tutorial explains attendees how to apply process mining to sensor logs acquired from smart spaces. Keywords. BPM, process mining, smart spaces 1. Introduction The aim of a smart space is providing people with automatic or semi-automatic services realizing the concept of ambient intelligence (AmI). The input for these intelligent services is represented by a sensor log, which is a sequence of measurement values acquired from sensors deployed across the monitored space. Many approaches have been proposed in the literature to automatically analyze sensor logs at runtime to understand the current context and to make decisions based on user preferences and habits. All of these solutions are based on models that relate the output of the sensors during a (potentially very short) temporal window, to a specific contextual information that can be then employed to act or reason on the state of the environment. Models can be either manually defined (specification-based methods) or obtained through machine learning techniques (learning-based methods). In the first case, models are usually based on logic formalisms, relatively easy to read and validate (once the formalism is known to the reader), but their creation requires a heavy cost in terms of expert time. In the latter case, the model is automatically learned from a training set (whose labeling cost may vary according to the proposed solution) but employed formalism are usually taken from statistics, making them less immediate to understand. Another difference between the two approaches is that whereas specification-based methods use human actions as main modeling elements, learning-based ones directly refer to sensor measurements, thus losing the focus on human actions and making even more difficult to visually inspect and validate produced models. On the other hand, taking as input raw sensor measurements usually makes learning-based methods easier to apply in a practical context; whereas, in the vast majority of cases, specification-based methods do not face the problem of translating sensor measurements into actions. Intelligent Environments 2019 A. Muñoz et al. (Eds.) © 2019 The authors and IOS Press. This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0). doi:10.3233/AISE190012 5