Adapting Content Delivery to Limited Resources and Inferred User Interest

This paper discusses adaptation policies for information systemsthat are subject to dynamic and stochastic contexts such as mobileaccess to multimedia web sites. In our approach, adaptation agentsapply sequential decisional policies under uncertainty. We focus onthe modeling of such decisional processes depending on whether thecontext is fully or partially observable. Our case study is a moviebrowsing service in a mobile environment that we model by usingMarkov decision processes (MDPs) and partially observable MDP(POMDP). We derive adaptation policies for this service, that takeinto account the limited resources such as the network bandwidth. Wefurther refine these policies according to the partially observableusers' interest level estimated from implicit feedback. Ourtheoretical models are validated through numerous simulations