Where are you? I'll ask your friends

An article in New Scientist explains how users of social networks might "unwittingly leak" location information through their friends.

A system developed by Adam Sadilek of the University of Rochester in New York predicts Twitter users' locations by looking at their and their friends' tweets. It does so using a dynamic Bayesian network, a way of encoding probabilities using graph theory that provides "a natural tool for dealing with two problems that occur throughout applied mathematics and engineering - uncertainty and complexity".

A trial using four million tweets from users in Los Angeles and New York was conducted and this found that:

a couple of weeks of location data on an individual, combined with location data from their two most sharing friends, is enough to place that person within a 100-metre radius with 77 per cent accuracy. That rises to nearly 85 per cent when you combine information from nine friends. Even someone who has never shared their location can be pinpointed with 47 per cent accuracy from information available from two friends, rising to 57 per cent with nine.

New Scientist mentions the obvious privacy concerns and point out some potential benefits in epidemiology, or to find nearby friends in case of a suicide attempt. It is worth reading the whole article, the rest of which discusses how a system might identify someone - or not - through their ability to identify photos of their friends, and a new programming system developed at MIT that "automatically enforces privacy policies".

New Scientist: What your online friends reveal about where you are (25 January 2012).