Please activate JavaScript!
Please install Adobe Flash Player, click here for download

uni'wissen 01-2016_ENG

Since Twitter messages can be collected by key- words and users, it’s easy to follow how they spread. In addition, it’s clear for every user who is following whom, how many followers each user has, and how they are connected with each other. Facebook, on the other hand, is much less accessible from outside. The researchers come up against a brick wall as soon as someone sets their account to “private.” “We don’t have access to profiles set to ‘private’ on Twitter either, but this is less common on Twitter.” Retracing the Path of a Message So Twitter it is. This social network has a “retweet” function for forwarding a message one has received. It and all freely viewable connec- tions between users allow the researchers to follow the path of a message – in the best case all the way back to its source. The more times a message has been forwarded, the longer it takes. Once he has collected all the necessary data, Fischer can identify the most influential users for well over 90 percent of the cascading tweets and thus retrace the precise path of the message. In the remaining cases, factors other than retweet- ing play a role in the forwarding of messages, such as keyword searches – but it is not yet pos- sible to determine these factors with sufficient messages per second. Experts speak in such cases of an information cascade. This may be seen when a pop star tweets a piece of news into the world, such as the announcement of a new album or a pregnancy. The news is regis- tered and disseminated at lightning speed by the star’s fans and followers – but it doesn’t reach many people beyond this group. “We refer to this as a star,” says Fischer. “The message spreads from a central point – in this case the pop star – in one or, at most, two waves. Then, only a short time later, it all collapses again.” What is much more fascinating for the research- ers is information that spreads through the social networks over a longer period of time and in several waves. A good example of such process- es is chain messages. They travel through the Internet for weeks before finally winding up in the big data trash heap. Then someone suddenly finds one of these messages funny again and shares it – sometimes years later. As an essential precondition for conducting their research, Fischer and his colleagues need access to the shared data and information on how the individual users are connected with one another. The messaging service Twitter has proven to be particularly practical for this purpose. Type, send, share: Thousands of messages flow through social networks each second – up to 150,000 at peak periods. Photo: Syda Productions/Fotolia 1414