Attended RecSys 2011 in October this year. Here is the offcial proceedings: RecSys 2011 proceedings. Conference had many interesting papers and industrial track was also pretty good with presentations from Ebay, Facebook, Twitter, Pandora, and Netflix.
- Xavier Amatrian posted a nice summary of the conference: Xavier's notes on RecSys 2011.
- The best paper award went to a paper by Yehuda Koren and Joe Sill: OrdRec: An ordinal model for predicting personalized item rating distributions. This paper talks about using user feedback, which is given as ordinal ranks of items, in collaborative filtering techniques.
- A short paper/poster by Jian Wang, Badrul Sarwar and Neel Sundaresan won the best poster award: Utilizing related products for post-purchase recommendation in e-commerce. This paper talks about post-purchase recommendations at Ebay. In this paper, author describes utilizing product categories, content and behavioral similarities, and temporal aspects (whether the item is available or not) for recommending products after purchase.
- Daniel Tunkelang from LinkedIn gave a tutorial on Recommendations as a Conversation with the user.
- Workshop on Personalization in Mobile Applications was quite interesting. Christine Outram from MIT gave a talk on "The Copenhagen Wheel" during the workshop. Janne Lindqvist from Rutgers University gave an interesting talk on Foursquare "I'm the Mayor of My House". Hyun-Jun Kim from Samsung gave a talk on "EmoSens: Affective Entity Scoring, A Novel Service Recommendation Framework for Mobile Platform", which was about recognizing emotions from smart phone touch, accelerometer, gyroscope sensors and recommendations based on that.
- Facebook's Rajat Raina gave a talk on "Learning Relevance in a Large-scale Social Network". He talked about how relevant ads are shown to the users. Facebook uses Open Directory concepts to map users and ads into concept space and do ad matching.
- Neel Sundaresan from Ebay gave an interesting industrial keynote. Here are notes from Siah: http://openresearch.wordpress.com/2011/10/26/notes-from-neel-sundaresans-keynote-speech-at-recsys-2011/
- Twitter's Pankaj Gupta gave a talk on "Recommendations and Discovery at Twitter". He talked about edge predictions used in "Who to Follow". Twitter uses features such as personalized pagerank, the most-common neighbors, top followings, etc. to generate "Who to Follow" recommendations. Interestingly less than 50% of twitter follows come from 2nd degree edges.
- Oscar Celma (BMAT) and Paul Lamere (The Echo Nest) gave tutorial on Music Recommendation and Discovery Revisited
- Workshop on Recommender Systems and the Social Web was also very interesting.
- Here are slides from some of the presentations: http://www.slideshare.net/tag/recsys2011
- Tutorial slides are available here: RecSys 2011 tutorials