Personality Traits of Twitter Bullies Using Deep Learning
Document Type
Article
Publication Date
2022
Abstract
Research in the phycology domain pointed out a relationship between the personality traits and the ability of a person to bully or be bullied. This paper aims at analyzing the personality traits, in terms of the Big Five model (BFM), of cyberbullies on Twitter communities. To achieve our objective, two deep learning-based models were trained to detect the BFM traits from a text; the first of them is a multi-label CNN based classifier while the second includes five independent binary CNN classifiers (one for each trait). The trained models were applied to a tweeter dataset of cyberbullying people, in order to identify their BFM personality traits. Experimental results showed that the characteristics of cyberbullies are high neuroticism, high extroversion, and low agreeableness. Moreover, modelling the BFM personality prediction problem using five binary classifiers achieves higher accuracy than utilizing one multi-label classifier.
Recommended Citation
Hamdy, Abeer Dr. and Hesham, Alaa, "Personality Traits of Twitter Bullies Using Deep Learning" (2022). Computer Science. 69.
https://buescholar.bue.edu.eg/comp_sci/69