Social information filtering: algorithms for automating “word of mouth”

This paper describes social information filtering, a technique for making personalized recommendations from any type of database to a user based on similarities between that user's interest profile and the profiles of other users. In effect, it automates the 'word of mouth' process by which people rely on the opinions of others with similar tastes. The authors implemented these ideas in a networked system called Ringo, which makes personalized recommendations for music albums and artists, and whose database of users and artists grows dynamically as more people use the system and enter more information.

The work tested and compared four different algorithms for making recommendations using social information filtering, and presents both quantitative and qualitative results obtained from use of Ringo by more than 2000 people. This early study of collaborative recommendation helped establish the foundations of what became collaborative filtering and recommender systems, demonstrating that automated, profile-similarity-based recommendation could work at scale with real users.