Context aware decision system in a smart home : knowledge representation and decision making using uncertain contextual information

Abstract : This research addresses the issue of building home automation systems reactive to voice for improved comfort and autonomy at home. The paper presents a complete framework that acquires data from sensors and interprets them, by means of IA techniques, to provide contextual information for decision making. The system uses a two-level ontology to represent the different concepts handled during the processing which also contains SWRL instances to automatise some of the reasoning. The focus of this paper is on the relationship between the knowledge representation and the decision process which uses a dedicated Markov Logic Network approach to benefit from the formal logical definition of decision rules as well as the ability to handle uncertain facts inferred from real data. The entire approach is situated wrt the Sweethome project whose aim is to make possible context-aware voice command at home.