Huberman et al 2008

Citation

Bernardo A. Huberman, Daniel M. Romero, Fang Wu1; Social networks that matter : Twitter under the microscope; 2008

author = {Bernardo A. Huberman
          Daniel M. Romero
          Fang Wu1},	
title  = {Social networks that matter: Twitter under the microscope},
year   = {2008},
url    = {http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1313405# },

Online version

[1]

Summary

The authors of this paper consider the differences in dynamics between the explicitly named follower/followee links found between users of Twitter, and the subset of those relationships that they refer to as real "friends." A "friend" is defined as a person to whom the user has directed at least two tweets.

The paper continues to compare the effect of growth of both the entire follower/followee network, and of growth of the "friends" sub-network. They find that there is a point at which a larger number of followers does not motivate a higher rate of tweeting. However, increases in the number of "friends" continues to increase a user's rate of tweeting, with no point of saturation. The authors cite this as evidence that, from an interaction point of view, the meaningful links in Twitter are only a subset of the links, and that these links must be detected through mutual interaction, rather than explicit follower/followee links.

Brief description of the method

The data set used consisted of 309,740 users. On average each tweeted 255 times, had 85 followers, and followed 80 other users. To find real "friends", the authors used ”@mentions” - the use of an @-symbol before a username, which directs the tweet to the mentioned user. Around 25.4% of all tweets in the dataset were directed.

The authors define $\delta$ as the ration of friends to followees for each user. It was found that most users have a $\delta$ value less than 0.1, an average $\delta$ value of 0.13, and a median $\delta$ value of 0.04. They claim that "this indicates that the number of friends users have is very small compared to the number of people they actually follow. Thus, even though users declare that they follow many people using Twitter, they only keep in touch with a small number of them. Hence, while the social network created by the declared followers and followees appears to be very dense, in reality the more influential network of friends suggests that the social network is sparse." 5

Datasets

The authors use the Twitter API.