A study on constructing dynamic context models for smart homes

Abstract

This paper discusses a method of constructing a context model, adjusting to environment and user preferences through learning, and updating the context model. An ontology representation model is chosen as a base for a context model in defining the relationship among concepts. Personal ontology is kept as the foundation for dynamically changing context models. Ontology reasoning is a technique used in determining the possible behaviors that the users are involved in by looking up ontologies and processing sensor data. Through filtering relative information, a more accurate behavior from a set of possible behaviors can be determined. Lastly, the correct behaviors are associated with context models according to an evolving process that lets subtle information that are learnt in determining behaviors.