Learning sparse heterogeneous user preferences
Models of user preferences will be at the core of the next generation of personalized Infor- mation Systems. We propose HPREF, an algorithm for learning a hierarchical, probabilistic preference model that integrates sparse preferences from multiple like-minded users in a principled fashion. Our preliminary experiments indicate that HPREF outperforms previous preference learning approaches and suggest several directions for further improvement.
|22nd Workshop on Information Technologies and Systems, WITS 2012|
|Organisation||Erasmus University Rotterdam|
Peters, M, & Ketter, W. (2012). Learning sparse heterogeneous user preferences. Presented at the 22nd Workshop on Information Technologies and Systems, WITS 2012. Retrieved from http://hdl.handle.net/1765/86055