Summer 2017
Analysis of Tagging on Flickr

Three Approaches for Collaborative Filtering Recommendation Systems
Tyler Nowicki
University of Waterloo
December 16, 2008

Collaborative filtering is a cooperative game. Each user of the collaborative filtering system has an agent which learns about the user by collecting feedback and observing the users interactions with the system. The agents in the system cooperate to generate a customized list of items for each user. A recommendation system is specifically concerned with finding items that the user will find interesting. To make the recommendations the agent must be able to predict its user's level of interest in an item. By cooperating, the agents are able to improve their individual predictions.

Download PDF