SINAI at SemEval-2023 Task 10: Leveraging Emotions, Sentiments, and Irony Knowledge for Explainable Detection of Online Sexism

Proceedings of the The 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 986–994 July 13-14, 2023 ©2023 Association for Computational Linguistics SINAI at SemEval-2023 Task 10: Leveraging Emotions, Sentiments, and Irony Knowledge for Explainable Detection of Online Sexism M. Estrella Vallecillo-Rodríguez, Flor Miriam Plaza-del-Arco L. Alfonso Ureña-López, M. Teresa Martín-Valdivia Department of Computer Science, Advanced Studies Center in ICT (CEATIC) Universidad de Jaén, Campus Las Lagunillas, 23071, Jaén, Spain {mevallecillo, fmplaza, laurena, maite}@ujaen.es Abstract This paper describes the participation of SINAI research team in the Explainable Detection of Online Sexism (EDOS) Shared Task at Se- mEval 2023. Specifically, we participate in subtask A (binary sexism detection), subtask B (category of sexism), and subtask C (fine- grained vector of sexism). For the three sub- tasks, we propose a system that integrates in- formation related to emotions, sentiments, and irony in order to check whether these features help detect sexism content. Our team ranked 46th in subtask A, 37th in subtask B, and 29th in subtask C, achieving 0.8245, 0.6043, and 0.4376 of macro f1-score, respectively, among the participants. 1 Introduction Sexism is defined by The Oxford Dictionary as prejudice, stereotyping, or discrimination, typically against women, on the basis of sex1. Sexism is in our daily life when people underestimate the opinions expressed by women, in conversations both oral or written that contains sayings and fixed expressions. Nowadays, with social media plat- forms, sexist comments are very frequent and they are widespread quickly promoting more sexist re- actions. Furthermore, sexism can be expressed in different forms, making detecting sexist comments a difficult task. Due to this reason, several shared tasks and many academic events are organized by the sci- entific community. For example, HateEval (Basile et al., 2019) is based on hate speech detection against women and immigrants. Other tasks such as EVALITA (Fersini et al., 2018a) or the task on Automatic Misogyny Identification (AMI) (Fersini et al., 2018b) are focused on misogyny detection. However, sexism is not only misogyny (violence against women) which detection consists of a bi- nary classification task, but also contains different 1https://www.oed.com/ types of sexism depending on how it is expressed, making the detection of sexism a multi-class classi- fication task. For this reason, there are shared tasks such as EXIST (Rodríguez-Sánchez et al., 2022) that promote the development of tools to detect sex- ism and categorize it according to the facet of the women that are undermined or the EDOS shared task. This paper describes the participation of SINAI in the Explainable Detection of Online Sexism (EDOS) shared task (Kirk et al., 2023) aimed at the identification of sexist language and its cate- gories. The main purpose of this shared task is to promote the development of different English language models to detect sexism and the explain- ability of this phenomenon. For this purpose, the organizers proposed three different subtasks. Sub- task A: Binary Sexism Detection is related to identi- fying if a comment is sexist or not. Once a post has been classified as sexist, …

Content truncated to 500 words