New Amazon Patents on Content Personalization

theodp writes "Defending its decision to concoct recommendations to steer customers to buy items at Amazon's new Apparel Store, a spokeswoman said Amazon "felt it would be evident to people that since the store was so new, we wouldn't have the transaction history to create database similarities." But in this just-published patent applicaton, Amazon earlier told the USPTO it's able to use product viewing histories to determine the similarity or relatedness between products for which little or no purchase history data exists. So which claim should you believe?"

Comments

I think you're mistaken

[clambert]

Amazon does have the technology to recommend items without customers having bought them in the past. They use product viewing history to do this.

In a new store such as Amazon's Apparal section, they don't have the transaction history OR the product viewing history, so they can't accurately recommend products to viewers.

I agreee that maybe they shouldn't be recommending products at all if they have no basis for their recommendations, but their patent doesn't conflict with their policy on Amazon Apparals. Basically, their patent is to allow them to recommend products to customers who haven't BOUGHT anything, but have just browsed through the store enough.

"People who wear clothes buy..."

[shoppa]

Whenever I've visited Amazon in the past month or so they've been pushing their apparel store on me, with recommendations wrapped in:

People who wear clothes buy...

Clearly, the've been able to assume from my past book purchases that I wear clothes! Quick, patent it!

Re:Yes, you can predict from books & cds to cl

[RebelTycoon]

Isn't it obvious that if I buy a book about God and Spirtuality that I am also interested in Anal Sex, which would mean that I am also interested in ...

Assless Leather Chaps