Improving bot detection of fake followers on Twitter via a hybrid B-HC optimisation algorithm

Alzeyadi H., DURAN F.

5th IEEE International Conference on Electrical, Computer and Communication Technologies, ICECCT 2023, Erode, Hindistan, 22 - 24 Şubat 2023 identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/icecct56650.2023.10179751
  • Basıldığı Şehir: Erode
  • Basıldığı Ülke: Hindistan
  • Anahtar Kelimeler: Fake accounts, Fake followers, Feature selection, Metaheuristic algorithm, Optimisation algorithm, Twitter
  • Gazi Üniversitesi Adresli: Evet

Özet

Sophisticated cyber threats are seen on Online Social Networks (OSNs) social media accounts automated to imitate human behaviours has an impactful effect on distorting public thoughts and opinions. OSNs are weaponized to diffuse deception, misinformation, and malicious activities, that forms a serious threat to society. The deceptive nature of imitating human behaviour has become a challenging and crucial task to detect automated accounts (socialbots). This research, however, proposes a hybrid metaheuristic optimisation algorithm for socialbot detection. Specifically, a hybrid B-Hill Climbing (B-HC) optimisation algorithm works in tandem with a k-NN nearest neighbour classifier to accurately select a relevant feature subset. It is applied to be tested for fake followers account on Twitter data. Experimental results showed that the proposed method is better than the traditional and the latest feature selection techniques as well as the rule-set methods. The B-HC alongside with k-NN method achieved promising results using only relevant feature subset.