Facing Uncertainty in Link Recommender Systems

This paper of Jean-Yves Delort and Bernadette Bouchon-Meunier talks about the uncertainty link recommender systems are facing.

More on the topic can be found @ http://www2002.org/CDROM/poster/63/

Dynamically Optimized Context in Recommender Systems

This paper by GhimEng Yap, AhHwee
Tan advocates context-awareness as a promising approach to enhance the performance of recommenders, and introduces a mechanism to realize this approach.

more on this can be found @ http://www.ntu.edu.sg/sce/labs/erlab/publications/papers/asahtan/context_mdm05.pdf

Movielens making good recommendation so far

In my 45 min interaction with movielens this week, I received good recommendation so far. There are lots of movies whose preview I have seen but did not get a chance to watch integrally. the rating movielens predicted for those movies is very close to what I expected from the previews. Some of the rating were a little off but most of them were pretty close.

I have rated 30+ movies so far but it just take too long to sort through the movies you have actually seen.

My next project is to pick one of the predicted good movie they suggested and go watch it to see if that will match my preference.

Will let you know.

Interaction Design for Recommender System

This paper by Kirsten Swearingen and Rashmi Sinha suggests the methodology for designing a good recommender. They studied 11 sites using recommender system and draw a good comparison on the approach user by each.

This is a good paper for this class as it demonstrates how to design a successful recommender system.

http://www.rashmisinha.com/articles/musicDIS.pdf

A trust-aware decentralized remcommender

Moleskiing.it is an information aggregator and adaptive web Recommender System. The high level goal of the system is to make ski mountaineering safer by exploiting information and communication technologies.

Precisely, Moleskiing is a catalogue of ski mountaineering routes in Trentino, Italy. Every route is identifiable by means of a unique URL. An user can create an identity on the moleskiing.it site and this allows her to keep an online diary (blog), her ``moleskine about skiing'' from which the site takes the name.

Source:
http://www.w3.org/2001/sw/Europe/events/foaf-galway/papers/fp/trust_aware_decentralized_recommender_system/

Rise of the Netflix Hackers

Confidentiality is still a big issue in recommendation. If you seriously think about it, the more they know about you, the better their recommendation will be. But the question is how much confidential informations one is willing to spare?

This article by Dave Demerjian tells how hackers are constantly working not only to gain more understanding of the systems but also how to compromise them.

http://www.wired.com/science/discoveries/news/2007/03/72963

Next Recommender Systems Conference

“The race to create a 'smart' Google”

Fortune magazine writer Jeffrey M. O'Brien, writes:

The Web, they say, is leaving the era of search and entering one of discovery. What's the difference? Search is what you do when you're looking for something. Discovery is when something wonderful that you didn't know existed, or didn't know how to ask for, finds you.



http://hci.epfl.ch/recsys08/