maandag 18 juni 2012

500 SA Journalists on Twitter


Are journalists on Twitter only following each other and preferable only those of the same media? That is an interesting question because the answer could shed some light on pluralism in the media and about the independence of journalists. The graph above may give an answer; it shows the network of Twitter relations between the top 500 South African Journalists on Twitter. These 506 tweeps share 2503 relationships; resulting in a density of 9%.  The colors show the different groups in the network and the size of the names reflects the authority of the tweep in the network.
Published at Journalism page Wits University, Joburg, SA: http://www.journalism.co.za/index.php/news-and-insight/insight/169-general/4991-sa-journalists-on-twitter-who-do-they-relate-to.html

Twitter network of 500 SA journalist. size set to authority and color to group

Elite
Because the density in the network is quite low, one cannot conclude that there is closely connected elite among the tweeps.  However if we look at various groups-that is tweeps how are closer connected- the picture changes. 

We see for example a small red group (2% of the tweeps) which is composed completely of the Caxton media; and bigger dark blue one of 11% dedicated to media24. The biggest, purple, group of 33% is the hard news media in South Africa.  The green group of 28% is a mixture of news and opinion and has the biggest number of freelance journalist. Finally we have a yellow group (18%) more dedicated to business/finance and IT, and the light blue group of 7% related to sports. Within these groups journalist follow each other more closely; that is the density in these groups are higher than in the overall network.

Authority is another characteristic in the network. Authority can be compared with Google Pagerank; the higher a page ends in a Google search the more imported the page is. Tweeps with a higher authority are for example more followed or re-tweeted.  They are more central persons or nodes in the network. Here are the top 20 of tweeps:
Top 20 Authority on Twitter
T_name
Medium
ferialhaffajee
City Press
nicdawes
Mail & Guardian
gussilber
Freelance
mandywiener
EWN
phillipdewet
Mail & Guardian
stephengrootes
EWN
maxdupreez

hartleyr
Sunday Times
adriaanbasson
City Press
702johnrobbie
Radio 702
bruceps
Business Day
verashni
Mail & Guardian
mandyldewaal
Daily Maverick
antonharber
Freelance
carienduplessis
City Press
art2gee
Freelance
guyberger
personal  on journalism
shapshak
Stuff
akianastasiou
Radio 702
brankobrkic
Daily Maverick

Compared to the result of a similar research by the British newspaper The Guardian (http://www.guardian.co.uk/news/datablog/2011/apr/11/journalists-twitter-following ), this South African picture is not much different. Journalists have a tendency to follow each other, in general, but more closely in groups. 

Data Journalism

This graph is an example of data journalism: finding a story in a pile of data. This new species in journalism is getting highly popular (http://memeburn.com/2012/03/data-journalism-where-coders-and-journos-meet/ ). Social network analysis like analyzing this twitter network is a special branch.  The central question in network analyses is: how many different walks you can make through a network of 500 persons, on the condition you visit them all? Many, of course, but some will be shorter than others; you will meet certain persons more than others (more central so more authority).  Social network analysis is a matter of mathematics (graph theory).  And the biggest problem is to give a meaningful interpretation to the numbers. The groups for example are constructed on basis of an algorithm that calculates a higher density. But the question is who these groups are? I have tried to give an interpretation on basis of the biggest number of media in each group.

Howto

How to make a graph like this? First you have to get the data. Luckily there is good and updated list of journalist on Twitter: Hacks List (http://hacks.mediahack.co.za/ ).  Now you have to scrape the data from the web page with Outwit Hub( a plugin for Firefox). Next clean-up the data with Google Refine. In the end you have a spreadsheet with twitter names, number of tweets and followers. Interesting but we need the relationships between them.  I used NodeXL, a template on Excel, to download and import the data from Twitter. Time for coffee, but this will take some hours. We can do the social network analysis with NodeXL, but the graphs are not so beautiful, and I like another program-Gephi-more. So Import the data in Gephi.
Within Gephi we make calculations for groups and for authority, with the build in algorithms. These are only numbers in spreadsheet. A picture, a graph, is more interesting. So we ask Gephi to draw a graph of the network where group is set to color and authority to size of the nodes.





Geen opmerkingen:

Een reactie plaatsen

Opmerking: alleen leden van deze blog kunnen een reactie plaatsen.