People who post personal information on Twitter take note: researchers say they’ve developed an algorithm that can piece together your life history via your Twitter stream.
Jiwei Li and Claire Cardie‘s technique analyzes tweets sent by a user and his or her followers, distinguishing between trivial events (“Homemade spelt granola for breakfast!”) and important ones (“Baby Lucia born at 3:05 this morning!”) to construct a timeline. The tricky part, they say, was coming up with a way to sort out the most substantial information automatically. As a solution, Li and Cardie divide tweets into four categories: personal and non-personal, and within each of those, time-specific and time-general. The most important tweets are personal, time-specific ones, such as the birth of a child or start of a new job.
To test their work, the researchers analyzed the Twitter streams of 20 regular users and 20 celebrities from 2011 to 2013. They asked the regular users to outline their own life histories, and used online biographies to outline celebrity lives. Then, they ran the algorithm against each Twitter account. Results were not perfect, but also “not bad.”
Of course, the whole process only works on Twitter users who post regularly and have enough followers for the algorithm to spot patterns in responses. And while Li and Cardie have no current plans to make their work widely available — for now — we can imagine a number of organizations from media to law enforcement agencies that may take a keen interest down the line.