Anomaly detection on social networks based on graph analysis and machine learning

The IJNCPS's Authors that presented the article:

  • mohammadreza Mohmmadrezaei 1Department of Computer, Ramhormoz Branch, Islamic Azad University, Ramhormoz, Iran
  • fatemeh peymani


A social network is defined as a social structure of individuals or organizations that interact directly or indirectly with relationships such as friendship, file sharing, and email. These include Facebook, Instagram, and Twitter. One of the concerns of users of these networks is an anomaly. Anomalies in online social networks can exhibit irregular and often illegal behaviour. These anomalies include malicious people such as spammers, sex hunters, account fraudsters, and online scammers who violate and abuse their privacy. This paper presents a new method for detecting a sample of anomalies called fake accounts on social networks. Given that these abnormalities cause disorder in the social network data, in the proposed method using network graph analysis and machine learning algorithms, the irregularity patterns are extracted, and then fake accounts are detected. The simulation results show that the accuracy and the error detection rate of the proposed method are improved compared to the previous methods.