Geolocating Twitter Users

Interesting research into figuring out where Twitter users are located, based on similar tweets from other users:

While geotags are the most definitive location information a tweet can have, tweets can also have plenty more salient information: hashtags, FourSquare check-ins, or text references to certain cities or states, to name a few. The authors of the paper created their algorithm by analyzing the content of tweets that did have geotags and then searching for similarities in content in tweets without geotags to assess where they might have originated from. Of a body of 1.5 million tweets, 90 percent were used to train the algorithm, and 10 percent were used to test it.

The paper.

Posted on March 26, 2014 at 1:10 PM • 12 Comments

Comments

David in TorontoMarch 26, 2014 4:09 PM

It appears to be two fixed simple substitution ciphers. Unless there is something else I'm missing, it's possible to break this by hand even without any message depth.

Given any message depth and computers ...

I wonder how secure they thought this was?

David in TorontoMarch 26, 2014 4:12 PM

Please ignore my last comment I clicked the comment link on the wrong article. Apologies all.

DBMarch 26, 2014 5:10 PM

By analyzing the commenters here, I figure one of them is in Toronto.

It is actually amazing how much data we give off about ourselves, that can be correlated and built up into a very detailed picture...

Tynhfdghtuc from VenusMarch 26, 2014 5:16 PM

And now that you have some more data, can you get more results ?

I'm curious :)

AnuraMarch 26, 2014 6:00 PM

Like, dude, no one can figure out where I am from my posts. I'm gonna to go grab a soda.

bob mcbobMarch 26, 2014 10:08 PM

Anura, you're by the soda machine! says Holmes the mysterious genius. And how do I know? I'll-never-te-ell!

EvanMarch 27, 2014 7:43 AM

This is interesting from a legal perspective, in light of the current Twitter crackdown in Turkey.

1. Would statistical geolocation be sufficient to obtain a subpoena against Twitter or a similar company for a particular user's information?

2. Given that Twitter is based in the US, could Turkey or another foreign government get such a subpoena via a US law enforcement agencies through existing formal or informal international agreements?

3. Would statistical identification (probably not limited to geolocation) be sufficient evidence to begin criminal proceedings or obtain a conviction?

Jonathan WilsonMarch 28, 2014 12:50 AM

Looking at who a given person follows (or is followed by) can help figure out where they are too, often people follow twitter accounts connected to people and entities (fast food joints, restaurants, media, transport, government agencies, local politicians, businesses, stores etc) from their area.

David in TorontoMarch 28, 2014 8:58 AM

Perhaps I am just saying I'm from Toronto. I might be from Cincinnati.

Besides even if I am from Toronto, there were 5 David's in my grade 5 class. So now you might guess my year of birth when David was a popular name.

Or I might be a German Sheppard named Dutchess. Possibly in Toronto. Or Cincinnati.

:)

David in TorontoMarch 28, 2014 9:02 AM

@ Tynhfdghtuc, Venus Texas perhaps?

BTW - how do you pronounce that?

MY Venusian is rusty but AFAIK n is pronounced like m and hfdghtuc is silent. Hi Tim. :)

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Resilient Systems, Inc.