Or, really, the fourth book, Breaking Dawn. Want to know why? Because I own all of Douglas Adams’s books*. Yip. Any clue what the connection could be? I’ve got one idea, but I’m not poisoning the discussion.
It specifically lists So Long, and Thanks for All the Fish and The Hitchhiker’s Guide to the Galaxy*, to be specific.
Breaking Dawn and So Long, and Thanks for All the Fish are both the forth books in their series?
They are both significantly worse than the previous books?
Maybe, but not what I was thinking. In fact, I just came up with another similarity.
But I’m only posting now to comment about how tired I must have been to misspell Twilight. :smack:
IMO their algorithm is crap - but the connection is, if I recall correctly, simply that others who bought these books, also bought those books. I get some pretty bizarre recommendations sometimes.
Ew, thanks for the warning… I’m pretty sure Amazon knows I have the Hitchhiker books, but it’s never recommended Twilight to me. Maybe it’s lurking in there somewhere, just hiding behind a few hundred textbooks, science fiction, and Discworld.
I work on software like this (not for Amazon but it’s similar). In most cases we never look at the attributes of the books (or shoes, or TV shows, or DVDs). The correlations are based on users preference. In all likelihood you were recommended Book A because there’s a statistical correlation between people who purchased that and Book B. It can lead to some interesting recommendations when you are talking about obscure titles, but in this case I bet there is a lot of data.
Retailers can make the algorithms fancier than that, but you end up working against your best interests. The results that produce the best ROI are to let the users’ actions speak for themselves.
This is the key: The point of a recommendation algorithm is to generate revenue for the organization running the algorithm, not to produce the best possible recommendations. It’s probably possible to do better than Amazon is doing right now, but the folks who find such an algorithm and figure out how to make it run sufficiently quickly on Amazon’s database will be producing doctoral dissertation-quality work.
People who enjoyed this thread might also enjoy the threads
Odd Amazon recommendations
I don’t need your commentary on my life choices, Amazon.com!
Most bizarre Amazon recommendation ever
Amazon recommendations that make no freaking sense
Check your bots, Amazon.
Amazon’s book recommendation engine is dumb – any better ones?
Funny, in the “Amazon recommendations that make no freaking sense” thread, the OP is asking about the connection between Pride & Prejudice and zombies, circa 2006.
Hey! I’ve got a thread going about this http://www.amazon.com/Pride-Prejudice-Zombies-Classic-Ultraviolent/dp/1594743347/ref=sr_1_1?ie=UTF8&qid=1248907422&sr=8-1
very thing: http://boards.straightdope.com/sdmb/showthread.php?t=526308&highlight=zombification
Nicely done. 
I don’t understand, the connection between Adams and the Twilight books is perfectly clear, what’s the problem here?
Adams writes comedy and the Twilight books are a joke.
Easy enough. Move along.