Warning: mysql_query(): Access denied for user 'memebawpmc'@'localhost' (using password: NO) in /usr/www/users/memebawpmc/wp-content/themes/NuMemeTheme/single.php on line 8 Warning: mysql_query(): A link to the server could not be established in /usr/www/users/memebawpmc/wp-content/themes/NuMemeTheme/single.php on line 8
In the last post of this series we looked at what the most popular hashtags in conversations around two of the most powerful tweeters in South Africa, Helen Zille and Jacob Zuma, revealed about the way politicians use Twitter.
Our next piece of research went one step further by mapping and visualizing a large portion of the actual network of connections the underlies the South African Twittersphere. In our network, people were connected if one person followed another person.
The way in which we collected our data ensured that it was biased towards collecting people with many connections rather than few connections. In this way, we were able to collect a network of over two million people stretching far beyond the borders of South Africa.
Using this vast network as a starting point, we narrowed our network down to a core of roughly 5 000 people who were all connected to at least one of our three accounts of interest, Helen Zille’s personal account (@helenzille) and Jacob Zuma’s personal (@SAPresident) and official (@PresidencyZA) accounts. In addition, we applied one final criterion to arrive at our network: a person had to have at least 10 followers. This removed many smaller and spam accounts.
The resulting network looked like this, where the size of each person’s node represents the number of followers that they have:
Note: The Malema account is a parody one. Interestingly, it has more followers than any of the others with the name of the former youth league president.
The top ten people in our network in terms of number of followers are:
Once we had this network, the first thing that we did was identify the sub-communities within each network. Using a modularity algorithm, five clear sub-communities were identified. We coloured each person in the network based on the community they fell into.
The below chart summarises the communities that we identified, with the size of each pie slice representing the size of each community. Unsurprisingly, it appears that most people in our network primarily follow sportsmen and celebrities, with news media coming in second:
Unpacking these communities, we find that the top 10 people in each community in terms of the number of followers that they have are:
As we can see, our algorithms have generally done a very good job of separating people out into homogenous sub-communities. This helps us to understand who the trend setters and influential individuals are that frame national discussions… at least on Twitter.
However, simply looking at who has the most followers does not take us far enough in terms of understanding real influence. To understand who is really influencing whom we need to dig a little deeper, which we will do in the next article…
It is unlikely that we collected every single person in South Africa that uses Twitter, however, it is clear that we were able to collect a wide cross-section