A few words about the timeline of the Libya conflict I’ve created. Click the image below to play around with it or keep reading for the behind-the-scenes story.
What I initially had in mind for my project was to create a web presentation that automatically generated a collage of the images (from Twitpic and Yfrog links) and videos (YouTube links) posted on Twitter with the #Feb17 hashtag (the Libya equivalent of #Jan25). I spent hours trying to get it to work in Flash, but the only Twitter element I was able to get automatically imported into my Flash stage was the text tweet. After seeking the help of several of CUNY’s Flash gurus, I concluded that Flash wasn’t suitable.
Then, with the help of Daniel Bachhuber, I moved on to JavaScript. It would probably do the trick, but then came another problem: The amount of unique Twitter content is remarkably little relative to the retweeted content – in other words, there’s a whole bunch of the same stuff. So if it was to work, I would need some rather advanced code to filter away the retweets (many retweets don’t use the ‘RT’ acronym, so merely filtering away anything containing ‘RT’ wouldn’t work).
So I had to ditch the idea of automatic curation. Daniel recommended me to take a look at ProPublica’s open-source timeline, TimelineSetter, a tool I had stumbled upon before, but never played with. I decided to use it to generate a timeline that would show the responses the major developments in the conflict got on social media.
My game plan then looked something like this:
- I identified the most important dates to plot into the timeline (using my own previous reporting on the conflict and text-based timelines from news agencies).
- I used Google Realtime Search to find the Top 5 tweets.
- I used YouTube, Twitpic and Yfrog’s embedding functions to add images and video. The images were found using Google realtime, by searching for e.g. ‘#Feb17 and twitpic’.
I chose Twitter and left out the Facebook content, mainly because it doesn’t generate feeds and because the content isn’t as quantifiable (e.g. for generating top 5 lists) as Twitter.
Update about the technical challenges with TimelineSetter coming up…
Thanks to Daniel Bachhuber for technical assistance and feedback on my failed attempts 😉
