Meta, Twitter, Spam Accounts, and Networks!
According to a report that was recently released by the Wall Street Journal, Meta and Twitter have decided to remove spam accounts that were found to be promoting Pro-Western content to Middle Eastern and Central Asian countries. Interestingly, the United States was assumed to be the “presumptive country of origin” for these spam accounts, with the culprit being either the U.S. government or one of its contractors.
In recent years, many social-media companies have begun to implement rigorous policies to take down accounts that send out spam messages in coordination with other accounts. Spam accounts have been enemy #1 of social-media platforms ever since the results of the 2016 presidential election, in which Russian fake accounts injected divisive political content into American social media feeds. But how exactly do social-media companies identify spam accounts? Here are some indications:
- Suspect account has a low followers/following ratio
- Suspect account has no mutuals with those it contacts
- Suspect account has little to no followers/following
From a networks perspective, we learned how to identify spam accounts using the Strong Triadic Closure property. For those who are not familiar, the Strong Triadic Closure property states that it is not possible for two nodes to have a strong relationship with a common node and not have some form of connection with each other. What is so interesting is how frequently spam accounts violate this property; Spam accounts tend to contact people who do not know each other. This violation makes it easier for social-media companies to identify them in their network. So interesting!
The Wall Street Journal Article: