Exploring the Formation of Collective Memory During the Tunisian Revolution : An AI-Driven Analysis of a Social Media Annotated Corpus

The following study is a pilot study in explorative research that aims to investigate 1) the relationship between Memory practices and emotions related to political events and 2) the extent of Artificial Intelligence (AI)- powered tools in facilitating the study of this relationship by conducting large-scale annotations. The research project that this pilot study takes part in focuses on the Tunisian Revolution that started in December 2010 – January 2011. This study constructs a corpus of Arabic tweets, creating the foundation for all and any future thematic and explorative studies. This study uses Chat GPT 4 as a sentiment and emotional tone annotation tool. We evaluate the tool's proficiency in determining sentiments and emotional tones in a monolingual corpus (Arabic) that contains standard Arabic as well as various dialects of Arabic. The findings of the study suggest that there has been a general disillusionment among all political figures in Tunisia post-2011. Despite the variety in the political orientations of all four personalities studied, the dominant sentiment has been negative throughout the years, highlighting dissatisfaction. The study also indicates that while Chat GPT is proficient in determining sentiments, it still lacks in identifying emotional tones. Moreover, the study confirms previous challenges and limitations to the tool that other studies have identified regarding corpus size and prompts.