I was reading an article from TechReview about Recommendation systems and it struck me that services such as Amazon's "Customers who bought this also..." are a lot like what we do when we do Readers Advisory. The difference is that Amazon and iTunes, and Netflix make recommendations using mathematical algorithms where we have traditionally relied on our own past readings. The goal is the same "based on what I am interested in now, predict what I may be interested in next"
Since we have a relatively small staff, maybe we should outsource a large portion of our reader's advisory work to Amazon recommendations?
The trick would be to refine the recommendation algorithm so that it can make good recommendations based on who we are (kid vs adult, mystery vs romance reader) and what we browsed vs bought.
Pandora and Slacker are internet music site that select and play music that they think you will like based on the name of a band or artists that you say you like already.
Maybe the first step to outsourcing/supporting readers advisory is to find the right recommender for each type of material. pandora for music, IMDB or rotten tomatoes for DVDS, and Amazon for books?
No comments:
Post a Comment