A traditional poll surveys a few thousand people over several days. With social media, we could engage millions of people in hours. What could we learn from the wisdom of the crowd?
So We monitor the 2012 Presidential election on Twitter. Currently, we are tracking:
We follow approximately ten million active American Twitter users. These users are filtered based on their locations and connections. Thus we could remove most irrelevant tweets from spam-bots or foreign countries. A large set of keywords is used to track election-related tweets, and as of Oct 9th, there are over 60 million tweets since Jan 1st, 2012. Of course, with such amount of data, there are a lot more analyses that could be done. Suggestions are most welcome.
We are grateful for people in the open source community who created great tools that made this project possible. Our appreciation goes to Ubuntu, OpenJDK-Java, Twitter4j, MySQL, Flask, d3js, Google chart tools, Bootstrap and many others we are not even aware of.