Friday May 19, 2006 - 12:00, NSH 3002
Title: Collaborative Adaptive User Profile with Implicit and Explicit User Feedback
Speaker: Yi Zhang

Abstract:

With the advance of research in information retrieval and ever increasing success of search techniques, IR research is now moving into a personalized scenario where a retrieval or filtering system maintains a separate user model for each user. In such a framework information delivered to the user can be automatically personalized and catered to individual user's information needs. However, a practical concern for such a personalized system is that any user new to the system must endure poor initial performance because there is insufficient information about that user. To help the system better serve a new user, we use explicit and implicit feedback from the user to build user specific models, and we use Bayesian hierarchical models to utilize information about existing users to learn the model for a new user. We analyze the adaptive performance of the model using two data sets gathered from user studies where users' interaction with a document, or implicit user feedback, were recorded along with explicit user feedback, such as relevance judgements. Our results are two fold: first, we demonstrate that hierarchial Bayesian modeling approach effectively trades off between shared and user-specific information, alleviating poor initial performance for each user. Second, we evaluate the benefits of utilizing implicit feedback in user profiling and find mixed results on different data sets.