P. Faratin, J. Wroclawski, G. Lee, and S. Parsons
The Personal Router is a mobile personal user agent whose task is to dynamically model the user, update its knowledge of a market of wireless service providers and select providers that satisfies the user’s expected preferences. The task of seamlessly managing the procurement and execution of short or long term connection for a mobile user is further complicated because mobile users performs multiple, concurrent and varied tasks in different locations and are reluctant to interact and provide subjective preference information to the agent. In this paper we present a detailed description and a formal model of the problem. We then show how the user modeling problem can be represented as a Markov Decision Process and suggest reinforcement learning and collaborative filtering as two candidate solution mechanisms for the information problem in the user modeling.