In the post-Snowden era, everyone is talking about the extent of Internet surveillance – however, Don Morrison, of the Norwegian Institute of Science and Technology in Trondheim, is taking a different “path” to “countermeasures”, developing a method that allows for monitoring censorship on the Internet, and in particular on services such as Twitter and the Chinese Weibo.
According to a report in New Scientist, Morrison created a virtual social network simulation in which two users are considered “connected” if one of their posts appears in the other’s timeline. The “pattern” of connections between users thus provides the “metadata” used to analyze online behavior.
Most social networks are made up of clusters of “communities,” the ties between which create distinctive structures. When Morrison simulated the activity of government censors, who deleted at least 10% of posts, the “shape” of these structures changed significantly, indicating a “distorted” and less well-connected network. This was most evident in cases where popular posts that had been replicated (via retweets, for example) were “targeted.”.
Morrison's system was able to "detect" with 85% accuracy when large-scale censorship was taking place. It was also possible to program the system so that, by detecting changes in the "pattern" of the network, it could send an update/trigger an alarm to activists and protesters, so that they would know when the authorities were interfering with their publications.
Morrison envisions an automatic online censorship detector: "What is the 'weather forecast' for censorship on these networks today? I can imagine that putting pressure on governments to limit censorship," he notes.
His work will be presented at the International Workshop on Data Mining in Social Networks in Taiwan in May.
Source: naftemporiki.gr
