Theoretical foundations for user-controlled forgetting in scrutable long term user models

Emerging technologies are making it feasible for people to capture large amounts of personal information that can support important aspects of their lives. When that information is stored in a persistent storage so that it can drive personalisation, it is called a user model. There has been little exploration of systematic approaches to enable people to gain control over such user models using forgetting mechanisms. This paper first presents reasons why such forgetting mechanisms are desirable. Then it analyses core forms of human forgetting as the basis for a theoretical model of forgetting in user models. Our key contribution is to establish theoretical foundations for the design of mechanisms and interfaces for forgetting in stores of personal information, with the goal of user control of these mechanisms, so that we can enable people to achieve a new form of control over their personal information and its use.