Wherever you are, IBM can find you – if you're tweeting, that is.
And we don't mean via simple location sharing, as IBM staff have apparently worked out methods to divine Twitter users' locations purely from analysing the content of multiple tweets and working stuff out, Sherlock Holmes style.
Specifically, a new paper (spotted by The Register) claims that IBM researchers can now pinpoint a Twitter user's home location (i.e. the place from where they send most of their tweets, presumably home) to a specific city with an accuracy of almost 70 per cent.
The location is revealed by using two algorithms, the first of which examines the volume of tweets made by any individual, and any external information, such as mentions of places and services – though the system can apparently still work even if locations aren't "explicitly" mentioned by the user.
The Register notes that the second algorithm then predicts locations "hierarchically using time zone, state or geographic region as the first level and city at the second level."
The paper used some 1.5 million tweets from almost 10,000 users, and is certainly an interesting exercise – but one which is bound to go down like a lead balloon with privacy rights activists.
IBM doesn't envisage the system being used for anything shady (naturally), citing an example use case of a journalist following an event on Twitter, who would be able to work out which tweets are coming from folks who are actually present at said event.
The study also states: "Our examination of the discriminative features used by our algorithms suggests strategies for users to employ if they wish to micro-blog publicly but not inadvertently reveal their location."
When all is said and done, though, if you really value your privacy above all else, you aren't posting to social networks at all...