New Web App Recommends Books Based on Your Tweets Posted on December 20, 2012 by socialmediamaven

nikki van winkle

Posted on December 20, 2012 by socialmediamaven via http://mashable.com/2012/12/19/bookrx/

Reading a book is a significant investment of time. That’s why it’s common to ask friends for recommendations when looking for interesting literature. If you haven’t said it yourself, you’ve probably heard someone say the common refrain: “Read any good books lately?”

That method has worked fairly well thus far, but an innovative new web appsources recommendations from the person who knows you best — you.

Simply insert your Twitter handle into BookRx, and seconds later the app produces a list of categories and specific books you might enjoy. The app, which was launched yesterday, is a product of Northwestern University’sKnight Lab. Shawn O’Banion, a third-year PhD student, worked with his professor, Larry Birnbaum, to create BookRx.

“Twitter is really unique because it’s a stream of consciousness for the user,” O’Banion tellsMashable.

“Twitter is really unique because it’s a stream of consciousness for the user,” O’Banion tellsMashable. ”Typically you’re projecting an image of yourself on Twitter with the things that you say; while that might not be your true self, it’s actually who you want to represent on social media.”

It works by comparing the words, hashtags and mentions in your tweets to those of a sample user with pre-determined reading preferences. The app’s creators used a secret data gathering technique to assess the sample and a machine learning algorithm to link tweeted words to specific book recommendations.

O’Banion recently used a similar method to create an app called TweetCast, which aimed to use tweets to predict votes in the U.S. presidential election.

For some, there is something innately unsettling about AI predictions. It is even more disturbing when the computer is accurate. Unlike sites like Amazon and Google, however, BookRx shows you the exact words you tweeted that led to its various recommendations.

“Typically when you see recommendations online, you’re not given any explanation or reason for the recommendation,” O’Banion says. “It’s sort of like a black box, and I think that’s why people get kind of creeped out by it, actually — because it’s not transparent.”

BookRx is a fun tool to try, whether or not you’re in the market for a new book to read. It’s interesting to see how accurately your tweets correlate to your reading preferences. Give it a try and be sure to let us know what you think in the comment section.

courtesy of mashable.com
www.nikkivanwinkle.com