Prompt Engineering for Identifying Sexism using GPT Mistral 7B

EXIST is an ongoing series of scientific events and collaborative tasks dedicated to identifying sexism in social networks. The goal of EXIST - in this case hosted at CLEF 2024 - is to encompass the full spectrum of sexist expressions, ranging from overt misogyny to more subtle forms that include implicit sexist behaviours. A binary classification is the first task. Systems must determine whether a particular tweet includes statements or actions that are sexist. In this paper, we discuss the application of a Mistral 7B model to address the task in the hard labelling setup for English and Spanish. Our approach leverages a Mistral 7B model along with a few-shot learning strategy and prompt engineering. Thanks to our approach, on the English test set, our best run achieved an F1 of 0.56, and on the Spanish test set, it achieved an F1 of 0.51. In the global ranking, our approach was able to obtain an F1 of 0.53. Our selected approach is able to outperform some of the baselines provided for the competition while outperforming other LLM-based approaches.