Monday, 20 December 2010

NCL Social Networks


The New City Landscape maps NCL based on twitter messages sent in urban areas represent only one aspect of the data that was collected. For the maps we were focusing on the location at the moment of sending the message.

Now we are looking into the social network of the same data set and visualise how users communicate and how they are interconnected. Using the @'s and RT's in te messages a network was constructed establishing links between users that send directed tweets or retweet someone else's message.

Very quickly these social networks become extremely complicated and computational intensive. The old MacBook here is struggling with the larger networks of London and New York, were there are 20000 or 40000 nodes respectively to be calculated.

However the smaller networks of Paris and Munich draw very nicely in Gephi, the open source graph visualisation software used to render the following images.

The data we are using here was collected over the period of one week and the selection criteria is the geo location. It is a 30km radius around an urban area. This forms the starting point of the network with all the users within this set as seed nodes. Who ever they were in touch with is then added as a secondary node drawing the edges between these nodes.

Munich NCL social network with one very big player standing out, ProSieben. This is one of the big German TV channels.

MUH_ncl_100505_test02_graph
Image by urbanTick / Social network of Munich twitter users corresponding to the Munich New City Landscape map.
The network is derived from @ and RT twitter messages.
This draws to 772 nodes and 1193 edges.


See the corresponding NCL map HERE.

Paris NCL social network
NCL-PRS_graph02
Image by urbanTick / Social network of Paris twitter users corresponding to the Paris New City Landscape map.
The network is derived from @ and RT twitter messages.
This draws to 4959 nodes and 7390 edges.


And a zoom into the Paris twitterNetwork map (bottom right), showing how the main nodes are linking down to smaller one of nodes.

NCL-PRS_graph03
Image by urbanTick / Zoom of social network of Paris twitter users corresponding to the Paris New City Landscape map.
The network is derived from @ and RT twitter messages.
This draws to 4959 nodes and 7390 edges.


See the corresponding NCL map HERE.

The NCL maps cover a lot more cities, see the world map HERE. As these are processed we'll be showing them here on the bog so stay tuned, some exciting network graphs will come up.

No comments: