Open Window is a web application designed and implimented by BoxUnfolded designer Charles Aweida in collaboration with Douge Roche and KeVon Ticer that allows people explore the world as Tweets are continuously pulled from the geographic region defined by the radar-like display. While incoming tweets can be filtered by keyword, users can also look at trending topics in major cities. This provides a simple way to curate and explore Twitter by keyword and geographic location.
Satisfying a curiousity to see what people are tweeting about around the world.
There are several applications and sites for seeing what is trending on Twitter and offer access agregated Twitter data. Out application is an interactive data visualization where you can zoom around the world and see what people are actually saying in the form of a tweet stream. There is something earily futuristic in that you really can see what people are saying all over and we tried to capture this mood in the technical looking UI.
Currently we are testing what data offers the most engaging user experience. While we may eventually shift towards storing our own Twitter data, allowing us to parse and store the data as we see fit, currently we are combining calls to the search and streaming APIs. This offers a combination of current tweets, recently popular tweets (< 7 day old), and currently trending topics. We are testing our application in person at the moment, but plan to collect analytics once the application is made live soon.
Some other ideas that we’ve considered include: tracing retweets geographically, exploring changes in trending tweets over time, having pre-defined keyword fitlers (for news, sports, etc.), and displaying more information in the tweet stream itself (number of retweets, time of tweet, number of followers, etc).
User Feedback & Iterations
- Not everyone realized they could pan and zoom since we had removed the interactors in our initial design.
- Tweets were best received in the stream in blocks of 5 or more (instead of a constant stream of 1 or 2 tweets every few seconds).
- The circle defining the geographic region was too large.
- More information should be availble in the tweet stream.
The radar visualization was too large and did not offer feedback when new tweets are pulled. I redesigned the radar display and animated the crosshairs outwards to offer feedback when tweets are being pulled from the search API on pan, zoom, and idle events. Below is our first design which relied on mouse gestures. The updated design has a series of pan/zoom controls.