BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Combining Collaborative Filtering with Meta Data for Scalable Reco
 mmendations - David Stern (Microsoft)
DTSTART:20090811T140000Z
DTEND:20090811T150000Z
UID:TALK19318@talks.cam.ac.uk
CONTACT:David MacKay
DESCRIPTION:I will present a probabilistic model for generating personalis
 ed recommendations of items to users of a web service. The system makes us
 e of content information in the form of user and item meta data in combina
 tion with collaborative filtering information from previous user behaviour
  in order to predict the value of an item for a user. Users and items are 
 represented by feature vectors which are mapped into a low-dimensional `tr
 ait space' in which similarity is measured in terms of inner products. The
  model can be trained from different types of feedback in order to learn u
 ser-item preferences. Here I present three alternatives: direct observatio
 n of an absolute rating each user gives to some items\, observation of a b
 inary preference (like/ don't like) and observation of a set of ordinal ra
 tings on a user-specific scale. Efficient inference is achieved by approxi
 mate message passing involving a combination of Expectation Propagation (E
 P) and Variational Message Passing. I also include a dynamics model which 
 allows an items popularity\, a user's taste or a user's personal rating sc
 ale to drift over time. 
LOCATION:TCM Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR