In necessariis unitas, in dubiis libertas, in omnibus caritas
Image: apocalipse motorizado, the online presence of a transit policy demonstration group here in Sao Paulo.
As you may know, I have been learning how to perform graph analyses using a mixture of Gephi, Yed and Pajek.
Today’s modest breakthrough was discovering how to export a complete subnetwork — a partition — out of Gephi’s data laboratory into a CSV file compatible with the drawing tool yEd, the simplicity of whose output I appreciate.
Let us say that we want to write a story on the June demonstrations focused on public transit fare hikes in Sao Paulo and creating a more pedestrian- and bicycle-friendly city.
You have vaguely heard of a local chapter of Critical Mass, with whom you are familiar from New York. Your wife went on a few rides with them, in fact.
But this is not New York. How about trying to detect a local networked community to get you started?
Well, you have enough data on the Brazilian Web that you can use Pajek to generate several substantial k-neighbor rosters in relation to a prominent search engine result called Tarefa Zero, or “zero fare.”
This turns out to be the focus of a whole community of related blogs and other media which coincide with a fairly dense Gephi modularity class. We can separate it out as a coherent partition of the entire graph. We may also find relevant content elsewhere in the graph, so one chore we have is to visually inspect each partition, briefly. Thus, we discover The City Fix Brazil.
This partition is to be inspected and fiddled with, and then re-exported in CSV format for the XLS transformation needed to open the file in yEd.
The key to the confusion I ran into trying to accomplish this simple task: You must export the current workspace to another, vacant workspace inside Gephi. Now the unfiltered nodes and only the unfiltered nodes are present in the data laboratory. Export this datalab as a CSV or GraphML file.
Use the Export option in the main menu, not the export spreadsheet on the Data Lab tab.
The result, as drawn in yEd.
The heavy in-degree of the Folha de S. Paulo is interesting. We do not know what kind of interactions the transit-themed blog had with the paper and its reporters, but we can say that the Folha was very much in the critical limelight during these incidents and has take a combative stance against its critics.
Just ask the bloggers of Desculpa Nossa Falha, hit with a SLAPP suit — the Brazilians lack this term but ought to invest in it — for the sake of a pun — folha, “the page” and falha, “the failure.” It was argued that lampooning the paper’s style and content was an assault on its copyright. Zero Hora of Porto Alegre –“H-hour” — has to contend with Zero Fora, or “down with Zero Hora.” And so it goes.
As to the rest, I recognize Jornalistas.org.br, and that is about it — and the new Portuguese edition of El Pais, not to be forgotten. A memorial to the first movement for fare-free transportation in Joinville, Santa Catarina, is still on the air.
And so we conclude that somewhere out there are at least a 100 or so active bloggers passing along information — recycling it, as it were — who might be interested in … and interesting to … an interviewer.
Next steps: at the same time, in Gephi, you can apply the modularity filter to the subnetwork to see if there is any more significant internal structure.
In this case, the partition has three subdivisions: the fare protesters, a typical dense Global Voices node and a red cloud of blogs with no discernable point of commonality. This discovery adds a very interesting dimension to the story of the Sao Paulo protest movement. The Brazilian protests are no isolated case. An attempt is probably being made to reproduce a similar movement in Boston, which shows up in the graph along with zero-fare sites from various geographic locations.
And so the task was simple enough, although in the end I had to come across the solution by trial and error — Gephi users are a highly technically advanced clique except when it comes to technical English.
Meanwhile, Global Voices Online is everywhere, in every partition, it seems, as always. It has become the post-9/11Voice of America, a sort of shadow Crisis Group.
The GVO subset shown above, grouped by partition and mined from our network sample of 850,000 nodes, highlights the current activities of the group and its affiliates, most of them very long-term free and open software projects undergoing a rigorous development process.
This very post, by the way, was executed using the LibreOffice in various ways, from writing to formatting data. It handles hundreds of thousands of data points with poise when called upon to do so. Thank you.
Anyway, in this network of traffic bloggers and protest groups, as it turns out — partition 3 — there is a substantial cloud of blogging on public transit policy alternatives in English. Free Public Transit, for one, and Free Public Transports, but there are quite a few bearing the farefreecountryx.blogspot.com and FPT — free public transport — subdomains .
Consider the following:
Much of this hypnotic Etch-a-Sketch subnetwork consists of Indymedia sites — another constant source of hairballs in our social data! More data can be extracted by isolating and subjecting this tangle to a simple layout and filtering, in Gephi, even, if you prefer.
And so it goes. It would be very interesting to construct a timeline of the creation of Critical Mass and related groups in order to explore the diffusion in this subnetworked movement. A problem is the shortage of public libraries where one can pile through back editions of the major dailies on the movement. Not one of them, as far as I know, ever noted that this modest group of activist was plugged into a weakly centralized network of activists sharing both a discourse and a media strategy.
But mostly, I am proud of myself for having gotten this far in a fairly short amount of time with Gephi. Photoshop for social networks? Could be, although Photoshop is a more of a tool for altering data than of extracting it. It depends on how you use it.
Filed under: Brazil |