SussexAI at ArAIEval Shared Task: Mitigating Class Imbalance in Arabic Propaganda Detection
Year: 2024
Type: conference-paper
Abstract: In this paper, we are exploring mitigating class imbalance in Arabic propaganda detection.Given a multigenre text which could be a news paragraph or a tweet, the objective is to identify the propaganda technique employed in the text along with the exact span(s) where each technique occurs.We approach this task as a sequence tagging task.We utilise AraBERT for sequence classification and implement data augmentation and random truncation methods to mitigate the class imbalance within the dataset.We demonstrate the importance of considering macro-F1 as well as micro-F1 when evaluating classifier performance in this scenario.Our system scored 0.12 micro F1 score and ranked fifth among the participants.
Authors: Mary Fouad, Julie Weeds
Institutions: University of Sussex, Minia University
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4402671508