Abstract
Adaptation decisions made by context-aware applications on behalf of users are based on evaluations of current context and preferences of users. This context information is imperfect by nature and can cause applications to behave in ways that users do not expect. Applications that exhibit unwanted behaviour will negatively impact their usability and violate the trust users have in them. Intelligibility and control in applications can help users to understand why they decided to behave in certain ways, and to forgive the applications by enabling users to override the undesirable adaptation. This paper presents a non-monotonic rule based approach (defeasible logic) for modelling user preferences, which serves as the basis of decision-making of application adaptations. It facilitates automatic generation of explanations regarding reasoning of defeasible preferences. The model also supports creation of feedback mechanisms for nontechnical users to formulate their own preferences independently, and modify the adaptation decision process to control applicationbehaviours. Moreover, to demonstrate its applicabilitywe have designed a set of evolvable situations and a context model, which complement the defeasible preferences for building smart home applications to enhance health care of the elderly.