Analyzing gendered patterns in sentiments in comments under STEM YouTube channels

This is the authors' abstract. We don't add key points for this paper.

This study explores gender-based disparities in sentiments and discourse in comments on YouTube channels focused on science, technology, engineering, and mathematics (STEM). Although there have been considerable efforts toward achieving gender equality, biases and inequalities still persist in the representation of women in STEM, including in online communication spaces. A mixed-methods approach was used, based on a dataset of comments collected from over 100 curated STEM YouTube channels using the YouTube Data API. The comments were cleaned and preprocessed, and information regarding the gender of content creators was identified. The RoBERTa model was used for sentiment annotation, while Latent Dirichlet Allocation (LDA) was applied for topic modelling. The results show a clear difference in sentiment distribution between genders. Negative comments were more frequent in female-hosted channels and were often related to personal criticism, objectification, and appearance-based remarks, whereas comments on male-hosted channels were more neutral and focused on content. The findings indicate the presence of gender bias in online discussions related to STEM.