17 Şubat 2009

Review: "Social networks that matter: Twitter under the microscope"

In the paper, Huberman et al. made an interesting conclusion about social networks in general, giving specific metrics on a dataset from Twitter.

The main idea is that the networks that are seen on the popular social networking sites on the internet is exaggerated and unreal, at least when examined with communication aspects of the network in mind.

By looking at a relatively small portion of the Twitter accounts, the paper reached at some valuable observations.

For example, the paper notes that number of posts of a user increases with an increase in the number of followers, but eventually saturates due to the fact that the time a person can spend on communicative activities is limited by nature.

But when number of friends is plotted against the number of posts, we see the positive effects of directed attention, people tend to contribute more when they feel that they are followed by people they already care. Contrary to the previous diagram, the curve does not saturate as the number of friends increase.

One must note that the definition of "friend" plays an important role here, a user is called a friend when there are at least two replies between the two users. As you see, this is somewhat more stronger than the traditional definition of a friend in this type of networks which was described as if you follow a user and she follows back, you are friends.

Thus, the paper concludes that the social networks contain a hidden network embedded in the publicly declared connections. This means that to understand the dynamics of interaction across the networks better, one must try to separate this hidden network from the actual network seen at first glance.