User-centric inference based on history of context data in pervasive environments
Nikos Kalatzis, Ioanna Roussaki, Nicolas Liampotis, Maria Strimpakou, Carsten Pils · 2008
Pervasive computing systems need to be strongly proactive. Context-awareness contributes to this, thus minimizing human-machine interaction. Context-aware systems are greatly enhanced by the utilization of recorded history of the users' situations and interactions. In this paper, an approach is proposed for modelling, storing and exploiting history-of-context, in order to predict or estimate context information. The proposed framework is context-type-independent, requires minimal processing and storage resources, and can be used for data compression. It is based on multiple context prediction rule generation models, demonstrates high prediction success ratio, and has been empirically evaluated via extensive experiments.