Personal data capture and algorithmic modeling of user preferences and interests is maturing. While its effect is felt most strongly in commercial activities and marketing, long-term value to the individual and society will arise in other disciplines like education, cultural heritage, health, and social engagement. But longitudinal user interest modeling is challenging. It is also important to be able to share that model across multiple institutions and domains, even across country borders. And what is the role of the individual user? This talk explores some of the model-building considerations for longitudinal preference management, how adaptive interfaces might play a role, as well as the way that users could interact with a model at a particular time in an application, to reflect their immediate needs and context.

We will explore and discuss:

• Considerations for collecting interest/preference data, and its limitations
• Ways to think about evolving a model over time, so it remains longitudinally relevant
• Possible approaches to transparency and user control
• How to consider taking context into account, and have users refine what is important