Discovering Human Habits Through Process Mining: State of the Art and Research Challenges

A smart space is an environment outfitted with sensors and actuators that, starting from the raw measurements provided by the former, is able to provide services to humans, assisting them in performing daily life tasks and eventually making decisions based on user preferences. Unfortunately, the vast majority of algorithms in this field are supervised, thus requiring final users to conduct an annoying labeling effort of performed activities. In this chapter, we show how concepts and techniques borrowed from the area of process mining can be used to ease this operation. In particular, we propose a bottom-up discretization strategy to automatically segment a smart home log into meaningful portions called habits. We evaluate the proposed approach using a well-known dataset in the smart space literature.