TweetCatcha seeks to uncover the organic nature of news as it travels through Twitter over time, by examining the movement of NY Times articles through Twitter. Nick and me worked on this project for our data visualization class and my AS3 class.

The New York Times Newswire API is used to load news for the last 24 hours. The title and URL for the retrieved articles are used to search for tweets with the BackTweets API, a BackType service.

The articles are placed around the center arranged clockwise based on the time they were published. The tweets for each article emanate from the article near the center to the outside. These are based on the time difference between when the article was published and when it was tweeted. The 24 rings indicate the hour difference from 1 near the center to 24 near the outside.

Data was collected between November 13, 2009 and February 9, 2010 via a cron job set up to pull and store the data locally. The current database is 107 MB, with 15,327 NYTimes articles and 311,885 tweets for those articles.

Demo

Presentation:

The circular rings are rotated around a center point based on the time the news piece was posted. The size of the ring is based on the number of characters in the headline and different colors are used for each section (for purely aesthetic reasons). The dense area in the middle indicates that a majority of the headlines are around the same characters in length. Click on the image to view the visualization.

Here the length of the characters are represented as a equalizer-like histogram. Moving the mouse over the bars reveals the headline and generates/plays a pulse wave. The frequency is also based on the number of characters in each headline – low for shorter and high for longer headlines.

The mockup below is intended to be a generative tetris-like game. The pieces are generated based the headline that is displayed along with the piece. Each section is color coded and has a piece assigned to it. Higher points can be scored by attaching the pieces to headlines that are related based on terms or people mentioned.