Last year, I covered a number of the so-called “Twitter protests” in China (#cn220), Iran (#25bahman), and Algeria (#fev12). Since these protests began in January 2011, the Arab Spring has claimed many members of both ruling and revolting groups – Mubarak in Egypt, Gaddafi in Libya, Ben Ali in Tunisia, Saleh in Yemen, and countless civilian and military casualties throughout the region.
Despite this seemingly stark year-over-year contrast, some aspects remain the same. One year after the original #jan25 movement that began in Egypt, protesters again took to Twitter to air their discontent. In order to document this movement, I used my Python tweet archiving script to record all tweets over a window leading up to and after the January 25th event. The final dataset has the following properties:
- First tweet: 20 Jan 2012 02:20:58
- Last tweet: 27 Jan 2012 14:12:36
- 33,997 unique Twitter handles
- 193,935 tweets
To summarize these tweets, I produced the figures below to highlight who tweeted, when they tweeted, and how the tweets built user-to-user networks. The first plot below lists the 30 highest frequency users on the #jan25 tag.
This second figure displays the frequency of #jan25 tweets; each bar represents a five-minute interval, and the color of each bar indicates its intensity relative to other intervals in the sample.
This last figure is a network visualization of the 3-core subgraph of the giant component of the Twitter user mention/RT graph formed by this dataset. This simplified visualization of the graph contains 9,914 users and 62,631 edges, whereas the entire network is made up of 31,805 users and 89,554 edges.
While there are a number of trivial observations that could be stated about these three figures, I’ll refrain. Instead, I’ll be working on answering a more interesting question – how does one “movement” compare to another? Stay tuned for upcoming analysis ranking #jan25 against other Twitter movements.
Hi Michael, would it be possible to see the code for your second plot (tweets per 5 min over time)? Is it on github somewhere? thanks!
Hi Ben,
The code for the time series plot looks like the sample on the #cn220 post at the bottom.
This is a great example of how your Python script can work. I’ll stay tuned for your upcoming “Twitter movements” analysis posts.
Would love to see the code for the first and especially the third plot. Many thanks.