A computer scientist at Duquesne University is working to develop software that strips unique words and phrases from a piece writing to preserve anonymity for the author.
Each person has an individual style of speaking and writing, said Patrick Juola, one of the project’s lead researchers.
“Some people use bigger words than other people,” Juola said. “But it’s not necessarily a very good indicator by itself, partly because most people use words of about average length."
Juola, who directs Duquesne’s Evaluating Variations in Language Lab, said the software will take into consideration more than 1 million factors such as sentence structure and preferred prefixes. It will then analyze all these idiosyncrasies in tandem and remove identifiable features.
But who would need such a tool?
“If you were an anonymous whistle blower, like the person who wrote that op-ed for the New York Times about the resistance within the Trump Administration,” said Juola. “If we could have a genuinely anonymous way of communicating, then that would encourage people to come forward and increase corporate and governmental transparency.”
While it might protect whistleblowers, there may be more nefarious uses for the program.
“Possibly this could be used for plagiarism,” said undergrad researcher David Berdik. “Say I take a fellow student’s paper and want to submit it to a teacher or professor. I just run it through this program and it could change it up."
This project builds on Juola’s previous work: he also created software that identifies anonymous authors that was famously used to reveal that “Harry Potter” author J. K. Rowling also used the pseudonym Robert Galbraith.
The work is being done with a $300,000 grant from the National Science Foundation. Juola said he hopes to field test the new software in the next three years.
90.5 WESA receives funding from Duquesne University.