In 1997, when only 18 percent of Americans used the Internet, John Riedl had already become famous in the field of computer science.
The University of Minnesota professor had just finished creating an artificial intelligence media recommendation system that was considered ground breaking for its time.
âÄúWe were having a great time doing this research,âÄù Riedl said. âÄúWe were getting famous âÄî lots of people were finding it really interesting, and we started getting calls from companies.âÄù
Some of RiedlâÄôs first clients were E! Online and Amazon.com.
The recommendation system, called GroupLens, analyzed data from thousands of people around the world. Using that data, the system was able to figure out what each user was interested in based on the people they tended to agree with more, Riedl said.
In 1997, the system was expanded into a research lab at the University where students have built upon the algorithm Riedl invented for more than a decade.
âÄúIt was strange at the time to have a lab focused on [recommendation systems],âÄù said Aaron Halfaker, a graduate research assistant in RiedlâÄôs lab. âÄúNow there is a conference devoted to recommender systems.âÄù
Halfaker points to the success of the system on websites like Amazon.com, which eventually led to the financial backing to start GroupLens.
Fueled by numerous National Science Foundation  grants since 1997, the lab has quickly established itself on the Internet.
One of several creations from the GroupLens lab is MovieLens.org, where similar technology used for books on Amazon.com recommends movies based on 1,500 different characteristics such as violence or drug use.
MovieLens, also launched in 1997, had more than 50,000 users by September of the same year.
âÄúThe response has really been positive,âÄù GroupLens research assistant and University student Jesse Vig said.
An artificial intelligence algorithm goes out over the Internet and computes more than 14 million values that take into account the collective view of a certain film.
The algorithm also scans for certain keywords in text reviews from outside sources such as a movie review by a local newspaper or numerical ratings. It also feeds on reviews made by users of MovieLens.
âÄúThe qualities on MovieLens are like the position on a mixer,âÄù Vig said.
Vig metaphorically compared the system to a soundboard, and how a technician controls switches to obtain a certain sound quality.
âÄúWe have extended this metaphor onto the web âÄî helping people find things on the web and give people this degree of control,âÄù Vig said.
GroupLens is currently designing a similar Internet service called BookLens.
Funded by the Undergraduate Research Opportunities Program, the goal of the project is to have local libraries start using the web service to help recommend books to library patrons and book clubs.
âÄú[BookLens] will bring a lot of users together to share and talk about books,âÄù GroupLens undergraduate research assistant Michael Ludwig  said.
Local libraries currently do not have an adequate system for recommending books, Ludwig said. Some libraries rely on the technology Riedl helped establish on Amazon to recommend books, but AmazonâÄôs motivations differ from the goals of the library.
âÄúAmazonâÄôs motivations are all economical âÄî to try and sell books,âÄù Ludwig said.
Amazon also requires that the library link back to the online store and that the use of its technology be solely for commercial purposes, which doesnâÄôt align with the librariesâÄô motives.
GroupLens has already met with the Saint Paul Public Library to have the web service integrated into the system there, said John Larson, who is in charge of the St. Paul library website.
The integration of BookLens into libraries would enable the sharing of information with other libraries across a network, Ludwig said.
BookLens is non-commercial and is free to integrate, he said, adding that he hopes to release BookLens before he graduates in May.