I just launched a new project, BookRx, with Larry Birnbaum and the Knight Lab at Northwestern University which aims to provide personalized book recommendations to Twitter users. BookRx analyzes your tweets, specifically the words, hashtags, and users mentions, to rank categories of books and provide specific recommendations. It’s a fun way to find new books, explore your own tweeting behavior, and see how you match up to other readers on Twitter.
What does your Twitter activity say about who you will vote for? My latest project attempts to predict who you will vote for in the 2012 U.S. presidential election based on an analysis of your tweets. It examines the words, hashtags, websites, and usernames mentioned in your Twitter feed to predict if you will vote for Barack Obama or Mitt Romney.
This work has yielded some interesting distinctions in the topics and style of language used by supporters of the two candidates. Terms popular among Obama supporters include equal, women, register, cuts, and wealthy; while those common among Romney supporters are failed, government, leadership, obamacare, liberty, israel, and gas.
Further, it shows that certain major media outlets are indeed popular as information sources for either Republican or Democrat voters. For instance, Fox News is strong associated with Romney supporters while the The Huffington Post and CNN are popular among Obama supporters. See more in depth analysis and try your own username at:
I recently worked with Narrative Science to write personalized news stories based solely on Twitter data. The results were demonstrated at Fortune’s Brainstorm Tech Conference and the attendees. Check out an example story here.
I recently worked on a project which has been deployed as part of a large initiative to provide tools for Illinois voters ahead of the 2012 Illinois Congressional elections. It categorizes political communication on Twitter and provides visualizations to help voters understand the priorities of political candidates on important issues.
You are invited to complete a very quick survey which involves identifying and ordering your personal interests as it relates to news content. The survey should take no longer than 5 minutes and contains two questions in which you are asked to to rank your interests in terms of high-level news categories (e.g. Sports, Politics, Technology) and specific topics (e.g. Chicago Cubs, Elections 2012, Entrepreneurship).
The only requirement to participate in the survey is that you have a non-protected Twitter account, which you will be asked to provide. The data will be used for my graduate research and responses will be kept anonymous.
As part of my graduate research I’m building a tool that automatically analyzes what people are talking about on Twitter, as organized by your typical news categories (sports, technology, etc.).
Here’s a really useful and easy way to help me with this. Just click here, enter a username (don’t use your full name, please), and assign a category to each tweet that you see. If you revisit the site later, please use the same username.
Do as few or as many as you’d like, any help is appreciated.
Also, it looks great on your iPhone so make good use of that boring commute home or office meeting!