Project lead: Michael Fire, Washington Research Foundation Postdoctoral fellow
Collaborators: Dima Kagan, Aviad Elishar, Yuval Elovici
Winner of the 2016 “Best Commercialization/Translation Potential” prize at the WRF Perfect Pitch Session, Michael Fire and his collaborators are developing a new algorithm for identifying fake profiles on social media such as Twitter and Facebook.
Billions of people use online social networks on a daily basis, often sharing sensitive information such as home address, phone numbers, and locations. Fake profiles are used by criminals to obtain personal information which users unknowingly expose them to when connected. The criminals then capitalize on this data breach, resulting in a corrupt multi-million dollar industry.
Using a new unsupervised machine learning algorithm, Fire’s team can identify fake users based on the way the users are wired into the network, i.e. their connections to various communities. In the case of Facebook, profiles of users who were removed by their friends were used to construct a classifier that can locate fake profiles.