Business

Amazon snags BookMatcher patent

Amazon.com Inc. has been granted a patent on its BookMatcher service, a process that uses collaborative filtering to make book recommendations to patrons.

Written by Margaret Kane, Contributor July 28, 2000 at 12:47 p.m. PT

Amazon.com Inc. has been granted a patent on its BookMatcher service, a process that uses collaborative filtering to make book recommendations to patrons.

1 June 2000 - Patent No. 6,064,980, issued May 16 by the U.S. Patent and Trademark Office, covers a process by which products are recommended to a user based on the ratings given by previous users. The Amazon patent includes a technique to automatically update the database of recommended items based on the underlying database of books available. It also uses statistical analysis to validate recommendations before offering them.

Amazon's patents have been the source of some hot water for the company. After patenting its "one-click" ordering process, the online retailer filed suit against Barnesandnoble.com for infringement. That led to sharp criticism from many in the Internet community, and even prompted some users to boycott the service.

In March, Amazon CEO Jeff Bezos called for patent reform, saying the process needed to be overhauled to reflect the different nature of software innovation. Amazon did not return calls for comment on the BookMatcher patent.

"You're facing the issue where the patent office really doesn't have the ability to search these things. They just don't have the physical resources, and they don't necessarily have prior art documents," said Michael Bevilacqua, senior partner at Hale and Dorr in Boston. "These types will stand or fall in litigation."

Collaborative filtering is not new to the Internet or marketing. Patents dating back to September 1989 are referenced in Amazon's patent, titled "System and methods for collaborative recommendations." Amazon applied for its patent in March 1998.

MovieLens in focus
MovieLens was an early recommendation engine that used to collaborative filtering to predict what movies a user would like. The user rates a selection of movies, then the site compares that list to its existing database to find other possible matches. For example, if you rated three movies as top picks, it would scan its database for other users who gave similar ratings and see if those users had any other movies in common, then presents you with those results.

Barnesandnoble.com currently offers a cruder version of the recommendation engine based on the actual purchases of users.

John Riedl is chief scientist and co-founder of Net Perceptions, which developed the MovieLens site. He said that collaborative filtering is "a technology that's going to save the world for consumers."

"Consumers are often though of as these entities that are supposed be on the end of the firehouse of information that marketers want to present. This is technology that will help them find things that they want to buy," he said.

In other news, Amazon on Wednesday added digital audio books that customers can download to a computer or portable listening device.

The Seattle-based online retailer announced a deal with digital spoken word company Audible Inc., in which it has a 5 percent stake, to sell more than 4,700 audio books as well as comedy, newspapers, magazines and business information.

Visitors to the Amazon site will be able to purchase titles from popular authors such as Stephen King and John Grisham as well as classic literature from William Shakespeare and Mark Twain, the companies said. Customers can also subscribe to daily story selections from newspapers such as The New York Times and The Wall Street Journal, they said.

"Beyond immediate access to Audible content via streaming or downloading, our new Amazon.com customers will have the flexibility to listen at their PCs or take the audio programs away to enjoy in the car or while exercising," Audible Chairman Donald Katz said in a statement.

Reuters contributed to this report.