A preference modelling approach to support intelligibility in pervasive applications

Context-aware applications do not always adapt their behaviours in ways that users expect due to a variety of reasons. Applications that lack intelligibility are often incapable of offering explanations to users as to why they decided to adapt their behaviours in certain ways, and providing feedback mechanisms for users to take control of any unwanted adaptation. This can lead to loss of user trust, satisfaction and acceptance of these applications. However, providing intelligibility and user control in applications are non-trivial; it involves exposing internal working components that influence the adaptation decisions, rendering them understandable to non-technical users, and enabling user modifications to those components to correct unexpected adaptations. This paper describes a user preference model regarding application adaptations. The goal is to support intelligibility in applications by facilitating users in generation of clear mental models that enable them to understand the links between particular contextual situations and various adaptive actions. It also aims to support user control of application behaviours by assisting developers in the creation of appropriate feedback mechanisms. These are essential in preventing user frustration at erratic application behaviours.