Feature-Guided Automated Collaborative Filtering

Information filtering systems have traditionally relied on some formof content analysis of documents to represent a profile of user interests.Such content filtering is generally ineffective in domains withdiverse media types such as audio, video, and images, because machineanalysisof such media is hard.Recently, information filtering systems relying primarily on humanevaluations of documents have been built. Such automated collaborativefiltering systems work by discovering correlations...