6 Conclusion MRK97 Bradley N. Miller, John T. Riedl, and Joseph A an Experience with Grou- In this paper we out lined a framework for modelling Usenet useful again. In the collaborative filtering problem with PRMs. We USENIX r. 1997 Annual Technical model the Cf problem first using a standard PRM Conference, January 6-10, 1997. Ana then we extend model to account for hierarchical rela- heim, CA, USA, pages 219-233, Berkeley, tionships that are present in the data. hPRMs improve CA. USA. 1997. USENIX the expressiveness and context-sensitivity of standard PRMs, and also realize real-world performance bene UF98a L Ungar and D Foster. Clustering meth ods for collaborative filtering. 1998 [UF98b L. Ungar and D. Foster. A formal sta- Acknowledgements tistical approach to collaborative filtering 1998. and encouragement(as well as access to her sof Both authors thank lise getoor for useful discussion ware), Dale Schuurmans for his advice on this project and thank NSERC and the Alberta Ingenuity Cen- tre for Machine Learning for funding. Some of this work was done when RG was on sabbatical visiting CALD/CMU References Aha97 D. Aha. Special issue on "Lazy Learn- ing". Artificial Intelligence Review, 11(1 5),Fe BHK98 John S. Breese, David Heckerman, and Carl Kadie. Empirical analysis of predic tive algorithms for collaborative filtering In UA198, pages 43-52, 1998 [CHM97 David Maxwell Chickering, David Hecker man, and Christopher Meek. A Bayesian approach to learning Bayesian networks with local structure. In UA197, pages 80- Eac http://research.compaqcom/src/eachmovIe/. FGKP99 Nir Friedman, Lise Getoor, Daphne Koller, v earning probabilistic re- lational models. In IJCAI, pages 1300- [Geto2 L Getoor Learning Statistical Models from Relational Data. PhD thesis. Stanford Uni- KP98 D. Koller and A. Pfeffer. Probabilistic frame-based systems. In Proc. of the Fif- teenth National Conference on Artificial Intelligence, pages 580-587, Madison, WI lite McGraw-Hill. 1997.6 Conclusion In this paper we outlined a framework for modelling the collaborative filtering problem with PRMs. We model the CF problem first using a standard PRM, then we extend model to account for hierarchical rela￾tionships that are present in the data. hPRMs improve the expressiveness and context-sensitivity of standard PRMs, and also realize real-world performance bene- fits. Acknowledgements Both authors thank Lise Getoor for useful discussions and encouragement (as well as access to her soft￾ware), Dale Schuurmans for his advice on this project, and thank NSERC and the Alberta Ingenuity Cen￾tre for Machine Learning for funding. Some of this work was done when RG was on sabbatical visiting CALD/CMU. References [Aha97] D. Aha. Special issue on “Lazy Learn￾ing”. Artificial Intelligence Review, 11(1– 5), February 1997. [BHK98] John S. Breese, David Heckerman, and Carl Kadie. Empirical analysis of predic￾tive algorithms for collaborative filtering. In UAI98, pages 43–52, 1998. [CHM97] David Maxwell Chickering, David Hecker￾man, and Christopher Meek. A Bayesian approach to learning Bayesian networks with local structure. In UAI97, pages 80– 89, 1997. [Eac] http://research.compaq.com/SRC/eachmovie/. [FGKP99] Nir Friedman, Lise Getoor, Daphne Koller, and Avi Pfeffer. Learning probabilistic re￾lational models. In IJCAI, pages 1300– 1309, 1999. [Get02] L. Getoor. Learning Statistical Models from Relational Data. PhD thesis, Stanford Uni￾versity, 2002. [KP98] D. Koller and A. Pfeffer. Probabilistic frame-based systems. In Proc. of the Fif￾teenth National Conference on Artificial Intelligence, pages 580–587, Madison, WI, 1198. [Mit97] Tom M. Mitchell. Machine Learning. McGraw-Hill, 1997. [MRK97] Bradley N. Miller, John T. Riedl, and Joseph A. Konstan. Experience with Grou￾pLens: Making Usenet useful again. In USENIX, editor, 1997 Annual Technical Conference, January 6–10, 1997. Ana￾heim, CA, USA, pages 219–233, Berkeley, CA, USA, 1997. USENIX. [UF98a] L. Ungar and D. Foster. Clustering meth￾ods for collaborative filtering, 1998. [UF98b] L. Ungar and D. Foster. A formal sta￾tistical approach to collaborative filtering, 1998