You might think you’re being smart with your privacy by leaving the “location” field blank, but think again. New software has been developed that can discover your location based purely on the words you tweet.
Technology Review reports that researchers from Northwestern University and PARC have created an algorithm to detect where a Twitter user is from based purely on the tweets they send out – not based on the (lack of) location data they directly provide [PDF].
When you sign up for Twitter, you’re encouraged to fill in the “location” field. You can leave this blank, of course, and according to this research, one third of Twitter users do just this. Other users fill in the location field with things like “Justin Bieber’s heart” and “Earth”, which aren’t very enlightening for advertisers looking to target geographically.
Apparently, over 12 percent of users prefer inserting a pop culture reference into their “location” field rather than give away their actual location, 5 percent name a non-Earth location, and over 4 percent type something threatening to the reader. People really don’t want to give away their location on Twitter, it would seem.
But all that is moot now that this research paper has been published.
The researchers created an algorithm and applied it to 10,000 recent tweets from thousands of Twitter users. Looking only at their tweets (not their location information), they were able to discover the country and state that each user was from.
Some of the findings were obvious, like those who use the word “Colorado” most often are usually from Colorado. But there were some surprises in the data as well, such as those using the word “elk” being from Colorado as well.
This study has significant implications for privacy and security online. You are likely giving away much more information than you know in the way you tweet, the topics you choose to tweet about, the language you use, and even in the times you’re most active on Twitter.