Social network sites can predict links even between non-members
Want to stay away from Facebook, so as to safeguard your privacy? That may not work according to a study first published in April by a group of German researchers and reported in the science journal IEEE Spectrum.
According to the study, called ‘One Plus One Makes Three’ published by researchers at the University of Heidelberg, it is possible for a social network site to study the information available to them from members on the site, and use that information to predict the links between any two people, both of whom may not be on the site at all. The researchers used membership of Facebook among students at five universities from 2005.
The methods to make such predictions vary, but as one of the researchers, Katherine Anna Zweig, toldIEEE Spectrum by way of example: “if two nonmembers have at least five friends together on the social network platform side, and these five friends are also connected … then this is a very good indicator that the two nonmembers are also connected or are also friends.”
Zweig told Spectrum that the research was inspired by an incident in which a colleague of hers received an email from Facebook pointing out that he, as a non-member, might like to get in touch with a group of people whom the colleague knew, and who were already on Facebook.
In this case, when members signed onto Facebook, they gave the network access to their email address book, thus allowing the site to send such emails to all those on the email list. And while the researchers used the example of Facebook, they stressed that the research applied to other sites as well.
Researchers reached their conclusions using a limited set of data. But as they pointed out: “Social network platform operators, however, typically have access to much more detailed information …such as the age, sex and (approximate) location of their members; and if they provide messaging services they can infer the quality of an acquaintance from its communication pattern. Including such information into the features will likely improve prediction accuracy.”
Source :Economic Times