These tools may do the trick!
Now graduate students at Drexel University have released two potentially provocative stylometry tools, which could have larger repercussions for whistle-blowers, human rights advocates, hackers and, well, anyone who doesn’t want their writing traced back to them down the road. One tool helps identify the author of a disputed document, and another helps authors avoid detection. The students released early, “alpha” versions of their tools on Thursday at a convention of the Chaos Computer Club, a hackers’ group, in Berlin.
The tools, which are still young, imperfect and buggy, build on existing author recognition tools like Signature, a program created by Peter Millican of Oxford University, and the Java Graphical Authorship Attribution Program, or JGAAP, a similar program designed by Patrick Juola of Duquesne University.
The researchers said their recognition tool, JStylo, works best within a limited suspect pool (50 people or less), when there are 6,500 words of available writing samples per suspect (Tweets, e-mails and instant messages will do) and if the disputed document is 500 words or longer. Within those parameters, Drexel researchers said, they can identify an author “with a very high level of accuracy.”