Interpreting Information in Smart Environments with Social Patterns

Smart Environments (SEs) work in close interaction with their users. To perform properly, these systems need to process their information (both from sensing and to act) considering its meaning for people. For instance, when they manage workflows that represent users’ activities or consider the tradeoffs between alternative actions. Introducing social knowledge about the human context helps SEs to better interpret information, and thus people’ needs and actions. However, working with this knowledge faces several difficulties. Its level of abstraction is different from that directly related to system components. Moreover, it belongs to a background that is not frequent among engineers. In order to address these issues, this paper proposes the use of Social Context-Aware Assistants (SCAAs). These Multi-Agent Systems (MASs) manage explicitly social information using specifications conform to a domain-specific Modelling Language (ML). The ML aims at describing human aspects and their changes in a given context related to a SE. Social properties describe reusable knowledge using a template with these specifications and textual explanations. Working with the ML facilitates the semi-automated transformation of specifications to integrate social and other system information, derive new one, and check properties. Specific agents are responsible for managing information, and translating data from sensors to the ML, and from this to data for actuators. A case study on an alert system to monitor group activities, extended with social knowledge to interpret people’ behaviour, illustrates the approach.