Strings: A recommendation engine that’s transparent about its data mining

It’s no secret that companies like Amazon, Facebook and Google study our behavior to recommend products or push advertising our way.
A Bellevue, Washington-based company called Strings is trying to do the exactly that, but with very explicit user permission. Users can track the books they buy on Amazon, songs they enjoy on iTunes or other types of behavior on hundreds of other web sites. Combined with aggregated, anonymized data from other users, Strings can suggest products a person might like using a combination of techniques including collaborative filtering.
The product is designed to passively follow online activity like products you browse on Neiman Marcus’s Web site, provided you’ve activated Strings to track you on that property. You can fine-tune Strings to follow different kinds of behavior, whether it’s passive browsing or active behavior “starring”, “liking” or buying items. Users pick the online sites they want to monitor themselves on, and Strings doesn’t reveal information to advertisers. The company earns revenue through affiliate fees when it successfully recommends products that consumers later buy. Strings also has FriendFeed-style sharing, so you can send a feed of items you like to friends on the site. There are also granular privacy controls to manage who sees what.
“We believe that we’re an advocate for consumers. That’s precisely the reason we’re being upfront about tracking and why we offer people control over their data,” said Edward Balassanian, Strings’ chief executive and founder. The company has been self-funded to date.
Strings is part of a wave of companies like Blippy and Foursquare that try to incentivize and make users feel comfortable sharing their behavior. Provided you feel safe with this type of sharing, Strings might offer useful recommendations for books, music and all sorts of other products. If you’re privacy paranoid, then most online services have some sort of risk, and Strings tries to accommodate those concerns by offering detailed sharing controls.

Tags: collaborative filtering, recommendation, strings
People: edward balassanian