Gender Violence on Arabic Facebook: A Text Mining Study of the COVID-19 Era
Duha Alsmadi, Mohammad Emad Arafah, Abdulrahman Mohammed G. Habib · 2025
This study investigates social media discourse surrounding violence against females and minors during the COVID-19 pandemic through analysis of Arabic Facebook posts. Using a comprehensive text mining approach combining Latent Semantic Analysis (LSA), Latent Dirichlet Allocation (LDA), and emotion detection algorithms, we analyzed comments from three formal Jordanian news channels between 2020–2021. The study revealed distinct emotional patterns in discussions about violence, with female/minor-related content showing higher frequencies of sadness (62%) and fear (58%), while male-related content exhibited stronger associations with anger (45 %). Named Entity Recognition identified key societal factors including religious references, family relationships, and socioeconomic challenges. Statistical analysis showed no significant difference in sentiment distribution between gender-related discussions (K-S test p > 0.05). Our findings suggest that social media discourse reflects complex intersections between gender-based violence, socioeconomic pressures, and cultural factors during crisis periods.