| Título |
Multilingual Sexism Identification Using Contrastive Learning |
| Tipo |
Congreso |
| Sub-tipo |
Memoria |
| Descripción |
24th Working Notes of the Conference and Labs of the Evaluation Forum, CLEF-WN 2023 |
| Resumen |
We present our systems and findings for the Exist2023 (subtask 1), a shared task for multilingual sexism identification at CLEF 2023 [1]. Our system aims to accurately identify and evaluate the degree of sexism in social media content in a multilingual setting considering its subjective nature. We successfully integrated two variations of contrastive learning as an intermediate step in a conventional fine-tuning language model pipeline. Our approach not only outperformed the sole fine-tuned method but also achieved competitive results compared to the top scores in the competition. This substantiates the simplicity and benefits of our approach to the task of sexism identification. © 2023 Copyright for this paper by its authors. |
| Observaciones |
CEUR Workshop Proceedings, v. 3497 |
| Lugar |
Thessaloniki |
| País |
Afghanistan |
| No. de páginas |
855-861 |
| Vol. / Cap. |
|
| Inicio |
2023-09-18 |
| Fin |
2023-09-21 |
| ISBN/ISSN |
|
