2019/10/10 by Pulkit Parikh, Parikh, Pulkit, Harika Abburi +11 · 4 citations
Social Sciences · #Gender Studies in Language
paper · pdf · doi:10.48550/arxiv.1910.04602
Sexism, an injustice that subjects women and girls to enormous suffering,\nmanifests in blatant as well as subtle ways. In the wake of growing\ndocumentation of experiences of sexism on the web, the automatic categorization\nof accounts of sexism has the potential to assist social scientists and policy\nmakers in studying and countering sexism better. The existing work on sexism\nclassification, which is different from sexism detection, has certain\nlimitations in terms of the categories of sexism used and/or whether they can\nco-occur. To the best of our knowledge, this is the first work on the\nmulti-label classification of sexism of any kind(s), and we contribute the\nlargest dataset for sexism categorization. We develop a neural solution for\nthis multi-label classification that can combine sentence representations\nobtained using models such as BERT with distributional and linguistic word\nembeddings using a flexible, hierarchical architecture involving recurrent\ncomponents and optional convolutional ones. Further, we leverage unlabeled\naccounts of sexism to infuse domain-specific elements into our framework. The\nbest proposed method outperforms several deep learning as well as traditional\nmachine learning baselines by an appreciable margin.\n