Social media and other online platforms are frequently confronted with the problem of impostor accounts, e.g., individuals posing to be someone else on the platform. A conventional approach to identifying impersonators is to use a seed-set of verified and protected entities to perform a search in the social-media graph to discover other entities that are similar enough to a verified entity to potentially be impersonators of the verified entity. However, many platforms often do not have a list of verified entities available a priori. This disclosure describes the use of machine-learning techniques to detect impostors on social media and other online platforms in scenarios where a verified seed-list of authentic entities is unavailable. Additionally, the techniques discover new verified seeds, e.g., small-scale businesses, local organizations, etc. that can then be placed on the protection radar.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.