After seeing that twitter collects trends based on words frequently used in tweets, I wondered what those trends would look like over time. I believe that a visualization which animates the trends over the period of a year will serve both as a platform for nostalgia and an opportunity to show how the concerns of people are twitter may be skewed from those of people who do not use the platform.
The current implementation animates the most popular trends for each 24 hour period in increments of 12 hours.
As I work on this more I'm looking to incorporate feed from news services, additional data visualization for frequent trends, an interactive component that allows the user to follow particular trends of interest, and additional forms of trend word visualization.
The application is written in Flex/Actionscript with all controls custom made. The layout engine is extensible and will be the basis for testing how different layout algorithms may convey meaning to the trends over time.
The data is collected directly from twitter using a daily cron script written in python, and another script written in python which arranges the data in a compact CSV format for the application to use.