June 7 (UPI) -- Computer scientists at Edinburgh University in Scotland have built a new algorithm for detecting fake or dishonest online profiles. The new software is designed to pinpoint catfishes, users who create profiles with phony information about their age, gender, location and appearance.
Researchers tested their new computer on an adult content website where catfishes and fake profiles are especially common.
The algorithm analyzed 5,000 verified public profiles to train itself, using network activity and writing style to accurately estimate each user's age and gender. The model used connections between user behavior and personal information established in the learning phase to survey the data provided by unverified accounts and pick out incongruities -- those who are likely to be lying about their personal information.
The software proved roughly 40 percent of profiles on the site feature dishonest information. The analysis showed women were more likely to lie on their profiles than men.
"Adult websites are populated by users who claim to be other than who they are, so these are a perfect testing ground for techniques that identify catfishes," Walid Magdy, a researcher at Edinburgh's School of Informatics, said in a news release. "We hope that our development will lead to useful tools to flag dishonest users and keep social networks of all kinds safe."
Researchers are preparing to present their newest catfish-catching software at the International Conference on Advances in Social Networks Analysis and Mining, to be held late this summer in Sydney, Australia.