Carnegie Mellon University researchers have created an open source algorithm that can identify counterfeit social media accounts that attempt to gain community influence.
Fake Social Media Accounts Are Now Easier to Spot
The newly created algorithm helps to identify fake social media profiles that are used by criminals or spammers to gain influence online. The focus of the research team is to spot fraudsters in the presence of camouflage or hijacked accounts. The proposed algorithm named FRAUDAR is camouflage-resistant and works well with the available, tested data sets in the carried out experiments.
A Twitter experiment showed that FRAUDAR was able to identify more than 4000 counterfeit accounts in a social media graph consisting of 1.47 billion edges. Test results show that the novel method outperforms competitor tactics and detects injected fraud with a high accuracy. This is done even in situations where camouflage measures have been utilized by the criminals.
A lot of online sites and forums offer fake Facebook likes, Twitter followers, and fake review posting. This is done to mislead consumers about products and services, gain community influence and otherwise manipulate social media via different methods. The detection and neutralization of these activities are important for both the businesses who suffer bad reputation and the consumers who are led to false information.
A lot of fraudsters inflate reputations of their customers by initiating fake interactions or posting counterfeit reviews of products and services. The researchers formulated several of the characteristics that can be used to identify fraud accounts:
- Links on profile associated with malware or scams
- Clear bot-like behavior (e.g. replying to large numbers of Tweets with identical messages
- Account deleted
- Account suspended
You can read more about the test results and the algorithm itself by reading the research paper titled “FRAUDAR: Bounding Graph Fraud in the Face of Camouflage.”
