Multi-label Categorization of Accounts of Sexism using a Neural Framework

Abstract

Sexism, an injustice that subjects women and girls to enormous suffering, manifests in blatant as well as subtle ways. In the wake of growing documentation of experiences of sexism on the web, the automatic categorization of accounts of sexism has the potential to assist social scientists and policy makers in studying and countering sexism better. The existing work on sexism classification, which is different from sexism detection, has certain limitations in terms of the categories of sexism used and/or whether they can co-occur. To the best of our knowledge, this is the first work on the multi-label classification of sexism of any kind(s), and we contribute the largest dataset for sexism categorization. We develop a neural solution for this multi-label classification that can combine sentence representations obtained using models such as BERT with distributional and linguistic word embeddings using a flexible, hierarchical architecture involving recurrent components and optional convolutional ones. Further, we leverage unlabeled accounts of sexism to infuse domain-specific elements into our framework. The best proposed method outperforms several deep learning as well as traditional machine learning baselines by an appreciable margin.

Automatic Classification of Sexism in Social Networks: An Empirical Study on Twitter Data

“…Sharifirad and Jacovi [38] presented a categorization of sexism that included indirect, sexual, and physical sexism. A more recent study by [39] seeks to categorize accounts of sexism. Because the growing interest of hate detection towards women, other tasks to protect women from hate on the internet have emerged.…”

Space identification of sexual harassment reports with text mining

“…Topic modeling is a popular text mining technique to identify hidden semantic structure of corpora (Karami, 2015) and has been used for a wide range of applications such as analyzing social media data (Shaw & Karami, 2017; Karami, Dahl, Turner‐McGrievy, Kharrazi, & Shaw, 2018; Webb, Karami, & Kitzie, 2018; Karami, Webb, & Kitzie, 2018; Karami & Shaw, 2019), opinion mining (Hemsley, Erickson, Jarrahi, & Karami, 2020; Karami, Bennett, & He, 2018; Karami & Elkouri, 2019; Karami & Pendergraft, 2018; Karami, Shah, Vaezi, & Bansal, 2020), literature reviews (Karami, Lundy, Webb, & Dwivedi, 2020; Shin et al, 2019), spam detection (Karami & Zhou, 2014a), investigating social media strategies (Collins & Karami, 2018; Karami & Collins, 2018), and exploring medical and health documents (Karami & Gangopadhyay, 2014; Karami, Gangopadhyay, Zhou, & Kharrazi, 2015a, 2015b; Karami, Gangopadhyay, Zhou, & Kharrazi, 2018; Karami, Ghasemi, Sen, Moraes, & Shah, 2019). Closer to our work is a study that analyzed short‐length stories on the everyday sexism project website to classify sexual harassments to 14 categories (e.g., body shaming, rape, and threats) (Parikh et al, 2019).…”

#NotAWhore! A Computational Linguistic Perspective of Rape Culture and Victimization on Social Media

“…However, recent research has shown important sub-classifications of sexism that may be important for online media research. (Parikh et al, 2019) classifies tweets into 14 sub categories of sexism and we identify that not all 14 categories may have as drastic effects as victim blaming and slut shaming. Victim Blaming on online media directly leads to psychological disturbances for the victim and biased responses from authorities seeking legal action for such crimes (Gruenewald et al, 2004).…”