During the first UChicago Cognitive Computing Challenge, nine entries came from several departments and schools: Stats, Music Theory and History, University of Chicago Medicine, English Language and Literature, Harris School of Public Policy, Masters Program in Computer Science, and Chicago Booth.
CONGRATULATIONS BALLOT READY!
Ballot Ready wins the first Cognitive Computing Challenge! Aviva Rosman, a Harris School student, will receive $1000. Her entry and every entry submitted will be evaluated for consideration for a Spring 2016 Practicum in conjunction with the UChicago Masters Program in Computer Science. The practicum is a 10-week project with a student programmer and Cognitive Computing subject matter expert.
BALLOT READY'S WINNING ENTRY
Ballot Ready currently partners with universities and have their students manually research online and enter this information for us. The tasks they do are datamining (e.g. find the candidate website), determining what is a political stance (i.e. NOT “I support school” but yes to “I plan on closing 50 schools”), classifying that stance with the preselected political issues we have (i.e. “k12 education”), summarizing fluff to make it more readable for our users (e.g. “I have 3 beautiful children” becomes “she has 3 children”).
WHAT MAKES THIS HARD?
First, finding the data is difficult because of how disperse it is. Finding the list of candidates is itself hard, because roughly 30% of boards of election in the U.S. don’t have websites (source: http://www.techandciviclife.org/news/2015/11/5/illinoiscounties).
Beyond that, many candidates have their own website template, which makes them hard to scrape. Once we find the data, pulling out the information is timeconsuming. Finding which parts of the biographic information are relevant, and which parts of a statement on issues are actually stances. This is fairly time consuming when done manually, and it prone to bias and difficult to update regularly.
THE POSSIBLE SOLUTION
If we could automate the datafinding part, we could more easily catch everything relevant and we could update it more frequently. Ideally it would also be more cost-effective and faster, so we could cover more candidates running for office. Our goal is to provide information on every candidate on every ballot in the U.S.
As COO for BallotReady, I bring both familiarity with the problem and our proposed solutions. I joined the BallotReady team because of my passionate for informed democracy and my domain expertise in elections, campaigns, and government. Since beginning work on BallotReady, I have gained immense knowledge of the technical opportunities as well. As a result, I believe I am the right person to think through this area with a student programmer.