Giving the user explicit control over implicit personalisation

Personalisation has an important role to play in a pervasive environment, supporting resource management tasks and tailoring the system to behave in ways that suit the user. However, this depends on creating and maintaining an adequate set of information on the user's preferences. Experience has shown that one cannot simply rely on the user to input preferences, and it is essential to employ some form of machine learning to support this; on the other hand the user needs to be able to control this. Two major approaches used for this purpose are rulebased strategies and artificial neural networks (ANNs). Within the Persist pervasive system these two different approaches are being combined to give maximum benefit. However, in order to enable the user to maintain control over the resulting set of user preferences, it is essential that he/she can see the current state of this preference set and change it whenever this is required. This paper describes how this will be achieved.