Personis-LF: user modelling framework with user controlled forgetting for pervasive lifelong personalisation

Personalisation is at the core of the vision of pervasive computing. We envisage a lifelong user model as the driver of such personalisation. This would be the place that information about the user is accumulated from pervasive computing sensors and applications, so that it can be made available to applications as needed. One of the essential properties of such a lifelong user model is support for user control of the model and its use. Over time, a person’s model would accumulate considerable information. This paper presents a case for designing lifelong user models with intelligent mechanisms that enable a user to control forgetting within them. We present the arguments for supporting forgetting and link these to the design goals for mechanisms for forgetting. We analyze the nature and possible mechanisms of forgetting within a lifelong user model and provide a theoretical model as a foundation for forgetting mechanisms for long term user models.