EXPLORING VERSUS EXPLOITING WHEN LEARNING USER MODELS FOR TEXT RECOMMENDATION

The text recommendation task involves delivering sets of documents to users on the basis of user models. These models are improved over time , given feedback on the delivered documents. When selecting documents to recommend, a system faces an instance of the exploration/exploitati on tradeoff. whether to deliver documents about which there is little certainty, or those which are known to match the user model learned so far. In this paper, a simulation is constructed to investigate the ef fects of this tradeoff on the rate of learning user models, and the re sulting compositions of the sets of recommended documents, in particul ar World-Wide Web pages. Document selection strategies are developed w hich correspond to different points along the tradeoff. Using an explo itative strategy, our results show that simple preference functions ca n successfully be learned using a vector-space representation of a use r model in conjunction with a gradient descent algorithm, but that inc reasingly complex preference functions lead to a slowing down of the l earning process. Exploratory strategies are shown to increase the rate of user model acquisition at the expense of presenting users with sub optimal recommendations; in addition they adapt to user preference cha nges more rapidly than exploitative strategies. These simulated tests suggest an implementation for a simple control that is exposed to user s, allowing them to vary a system's document selection behavior depend ing on individual circumstances.