So this is an awesome little project I’ve only just heard of. Hands up if you’ve heard of Pandora, the music recommendation engine? If you haven’t and you live in Australia, then you probably can’t use it. Unless you decide to spoof your IP address to get access to it. Either way, Pandora’s Music Genome Project has delighted many for the past few years by predicting the kinds of music you will like based on the music you already enjoy. This sounds pretty standard for anyone used to using Apple’s Genius playlists on iTunes or Last.fm (or many other similar services). Where Pandora is different is that it uses computers (and people) to analyse the content of music for themes, pitch, tonality and a long list of other features to compile the ‘genome’ of a particular song, artist or album. Similar music recommendation services use an algorithm based on what other people like to recommend new stuff to you. Pandora uses an algorithm to create an on-the-fly radio station of songs similar to what you’re after based on what they actually contain (and not the ephemeral tastes of other people). If you haven’t tried it, trust me – it’s awesome.
BookLamp and the Book Genome Project are now doing essentially the same thing for books. Perhaps this sounds like a Skynet-style artificial intelligent takeover of our book-recommendation habits. That’s probably because it TOTALLY IS:
The key focus of the Book Genome Project is to use computer intelligence to extract and quantify, on a scene-by-scene basis, useful information about these key elements of books. Consequently, we created a “gene structure” for each of the three primary elements that we analyze.
The genomic analogy is imperfect but useful nevertheless: we defined the three elements of Language, Story, and Character as the literary equivalent of DNA and RNA classifications. Each gene category contains its own subset of measurements specific to its branch of the book genome structure.
Language DNA, for example, is made up of components that we call Pacing, Perspective, Description, Density, Motion, and Dialog, and each of these is an amalgamation of alleles which capture the expressions of different aspects of linguistic style.
They are using alleles. I mean, what is that even? Do you know? I don’t. I dropped science classes in Year 10.
At any rate, BookLamp has launched its consumer facing website and will soon be taking books on from publishers to add to their database. It’s not very big at the moment, and that affects the variety of books it can recommend (and the ultimate usefulness of the service), but when it gets up and running there’s no telling how good this engine will get. The best thing about it, from my perspective, is that it gets us out from under the yoke of customer-based recommendation engines like Amazon’s. Amazon’s service is great, but as I’ve written about before, it isn’t necessarily without bias.
What do you think of BookLamp’s offering? Will you be sticking to good old word-of-mouth, or will you give it a go? Sound off in the comments.