– Algorithm predicts your location in 24 hours with 20-metre accuracy (Wired, Aug 14, 2012):
A team from the University of Birmingham has won Nokia’s Mobile Data Challenge with an algorithm that can predict where you’ll be in 24 hours.
The software, developed by Mirco Musolesi, Manlio Domenico, and Antonio Lima, combines movement tracking with data from the people in your phone book. By looking at how someone’s movements correlate with those of other people that they know, the algorithm can make guesses about where that person is going, even if it’s a major change from their usual routine.
Previously, algorithms have successfully modelled an individual’s regular movement patterns, but haven’t coped well with breaks in that routine. By adding data from a user’s social graph, the Birmingham team was able to overcome this, yielding an impressive accuracy level of 20 metres or so.
Real-world applications for the technology could include tin-foil-hat scenarios like government location tracking, but it could also be a gift to advertisers — allowing them to serve up discounts for shops and restaurants in the places you’re likely to be. Furthermore, it could assist location-based app developers, by allowing them to make checkins easier for the user.
The biggest obstacle to the roll-out of this kind of technology is, of course, privacy. But Musolesi points out that some people are happy to share their location publicly. He told Slate that he merely finds it fascinating to uncover the hidden patterns that govern how people move about on a daily basis.