BehavE: Behaviour Understanding Through Automated Generation of Situation Models

Teodor Stoev, Kristina Y. Yordanova

Abstract: Automated systems for assisting persons to achieve their everyday tasks are gaining popularity, both in the application domains for supporting healthy persons, as well as for assisting people with impairments. The development of such assistive systems is a challenging task associated with a lot of time and effort and often requires the involvement of domain experts. To address this problem, different works have investigated the automated knowledge extraction and model generation for behaviour interpretation and assistance. Existing works, however, usually concentrate on one source of data for the task of automated knowledge generation, which could potentially result in simpler models that are unable to adequately support the person. To address this problem, in this work we present the BehavE methodology, which proposes the extraction of knowledge from different types of sources and its consolidation into a unified semantic model that is used for behaviour interpretation and generation of assistance strategies.