PanwarJayant at SemEval-2023 Task 10: Exploring the Effectiveness of Conventional Machine Learning Techniques for Online Sexism Detection

Proceedings of the The 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 1531–1536 July 13-14, 2023 ©2023 Association for Computational Linguistics PanwarJayant at SemEval-2023 Task 10: Exploring the Effectiveness of Conventional Machine Learning Techniques for Online Sexism Detection Jayant Panwar and Radhika Mamidi LTRC, International Institute of Information Technology, Hyderabad jayant.panwar@research.iiit.ac.in and radhika.mamidi@iiit.ac.in Abstract The rapid growth of online communication us- ing social media platforms has led to an in- crease in the presence of hate speech, especially in terms of sexist language online. The prolifer- ation of such hate speech has a significant im- pact on the mental health and well-being of the users and hence the need for automated systems to detect and filter such texts. In this study, we explore the effectiveness of conventional ma- chine learning techniques for detecting sexist text. We explore five conventional classifiers, namely, Logistic Regression, Decision Tree, XGBoost, Support Vector Machines, and Ran- dom Forest. The results show that different classifiers perform differently on each task due to their different inherent architectures which may be suited to a certain problem more. These models are trained on the shared task dataset, which includes both sexist and non-sexist texts. All in all, this study explores the potential of conventional machine learning techniques in detecting online sexist content. The results of this study highlight the strengths and weak- nesses of all classifiers with respect to all sub- tasks. The results of this study will be useful for researchers and practitioners interested in developing systems for detecting or filtering online hate speech. 1 Introduction The prevalence of sexist and offensive language online has severe consequences for the mental health and well-being of women. Emotional dis- tress caused by exposure to such language can lead to anger, sadness, and fear, which can negatively impact their overall well-being. Given the severe harm caused by sexist language online, it is impera- tive that automated systems be developed to detect and filter such texts. The shared task (see Kirk et al., 2023 for a detailed description) deals with the online detection of sexism in English and pro- vides the right opportunity to test the performance of automated systems in doing so effectively. The task comprises three subtasks: • Task A - Binary Sexism Detection: a binary classification task where systems have to pre- dict whether the text is sexist or not • Task B - Category of Sexism: four-class classification task where systems have to pre- dict sexist text into one of four categories: (1) threats, (2) derogation, (3) animosity, (4) prej- udiced discussions. • Task C - Fine-grained Vector of Sexism: an 11-class classification where systems have to predict sexist texts as one of 11 fine-grained vectors. In this study, we employ several different con- ventional machine classifiers like Random Forests, Decision Trees, etc. By conventional classifiers, we mean those that have been developed before the advent of more recent techniques like deep learning and transformer-based approaches. They are sim- pler in design, easier to interpret and understand, and …

Content truncated to 500 words