Sybil Detection in Online Social Networks
Hussain Ali, Ismaeel Malik, Saba Mahmood, Farah Akif, Javaria Amin · 2022
The internet has become a very important part of the society especially while socializing. Number of social media users are increasing at an exponential rate. While it has opened many ways for users to become fully connected with each other, it poses a threat where identify theft can cause damage to personal influence, reputation, well being, and wealth. The fake identities also referred as Sybils can manipulate reviews, followings, ratings of genuine users or they can deceive the community by portraying as an identity they are replicating. Identification of Sybils on social network is an issue and has been approached from different dimensions. We have however, proposed an Architecture that is based upon machine learning algorithms considering features from the social ids. The application developed on this architecture helps the user to identify given account as Sybil or genuine. The information from the social handle is fed into the trained machine learning model that predicts the user genuineness and its followers. The system is validated by checking 50 profiles of social identities. There were 20% Sybil ids and remaining were genuine ids. The proposed system accurately predicted the genuineness of the identities with 88% accuracy. Furthermore, the system also visually presented the presence of Sybil identities in the social graph of the user.