Improving Activity Classification Using Ontologies to Expand Features in Smart Environments

Activity recognition is a promising field of research aiming to develop solutions within smart environments to provide relevant solutions on ambient assisted living, among others. The process of activity recognition aims to recognize the actions and goals of one or more person in a environment with a set of sensors are deployed, basing on the sensor data stream that capture a series of observations of actions and environmental conditions. This contributions presents the initial results from a new methodology that considers the use of ontologies to expand the set of feature vector, which is computed by using the sensor data stream, that is used in the process of activity recognition by data-driven approaches. The obtained results indicates that the use of extended feature vectors provided by the use of ontology offers a better accuracy regarding the original feature vectors used in the process of activity recognition with different data-driven approaches.