Forum Moderators: open
..."The holy grail is to be able to capture all the customer's interactions in detail and get smarter about what not to recommend," said Usama Fayyad, chief executive and president of digiMine, the software company behind the online recommendations of J. Crew and Barnes & Noble. "We can recommend very well. Knowing when not to bother someone is much harder."...
..."To make interesting matches, a company needs a large number of people who have rated or purchased a large number of products. The cold start phenomenon arises when a Web site opens but neither of those criteria has been met. Joe Smith might buy the same book as Jane Doe, but without more data, it would be a stretch to predict that the next book Joe buys is one that Jane might want."..."The popularity effect is at work when results delivered by the computer are boring and obvious."...
(Personally, I'd call 'em recommendation systems, not recommender systems, but, like, whateverrrr... :) )
I've experienced both of the scenarios described at the beginning of the article in regards to Amazon. I guess it is a love-hate kind of deal but I wish more sites would offer something as advanced as what they have.
Human input is a necessity but it needs to be minimized so as not to take away the quickness and speed a user want when seeking the information or products they need. However, I think that clicking a few options here and there can save time in the long run for areas that are visited often.
On a side note, Google is maturing in their ad-words section, particularly with their content-targeted ads. I imagine sometime in the future that they'll play the recommendation-targeted area as well. I think we'll see the first of this through their partnership with Amazon.com
This is somewhat off the SEM path, but I wonder if an über-retailer like Amazon would ever try to lease aggregated customer purchase (or browsing) data to new retailers.
New retailers rolling out recommender systems would benefit by having sufficient data to overcome the "cold start" problem, and the über-retailer would benefit by gaining a new, reciprocal customer data stream...plus money, of course. The stream would be combined demographic and transactional data so that the new retailer could determine things like, "70% of people in Arizona who bought this CD browsed for these brands of clothing and also bought these books."
It would be along the lines of a search engine database provider such as Inktomi leasing its data to MSN, which is then free to do with it what it needs.
The whole thing would have to be handled very delicately, i.e., you'd want the data stream available for a limited period of time, contractual prevention of the data falling into the hands of other big-league competitors, and accompanying marketing materials assuring customers that their personal info wasn't being revealed.
Is this crazy? Is it already being done?
Nothing like a few good examples to get the discussion going, right? :)
>>Will Google will have access to Amazon's data?
Right now I think the ad partnership will apply to recommendations based on relations to the actual products being viewed. You look at a Beatles CD you see a ad from a site selling memorabilia. I can see Amazon studying that for awhile to see if it is profitable AND if it doesn't steal the user away. No sense in having an ad displayed that says %40 off all Beatle's albums.
If profits do pan out this way, I wouldn't be surprised to see Amazon and Google take it to another level where ads are served based on user habits and purchases. I don't think Amazon would have to give Google much information to carry this out.
On the leasing out aspect to other sites, I don't ever see Amazon going that far. They are setting themselves up to be the Walmart of the Internet so wherever they see high interests in Amazon, Alexa, and Google Ad-words by recommendation features, they'll probably just integrate that in themselves. If it is worthwhile.
Google on the other hand might since they purchased Outride, a personalized recommendation service. All sorts of nifty tools in their pockets. I think when they have all this content-product targeted ad-words figured out they will start focusing on narrowing down the targets. They'll go back to their partners with a whole new suite of tools to increase revenue for everyone.
