AraProp at WANLP 2022 Shared Task: Leveraging Pre-Trained Language Models for Arabic Propaganda Detection

Abstract

This paper presents our approach taken for the shared task on Propaganda Detection in Arabic at the Seventh Arabic Natural Language Processing Workshop (WANLP 2022). We participated in Sub-task 1, where the text of a tweet is provided, and the goal is to identify the different propaganda techniques used in it. This problem belongs to multi-label classification. For our solution, we leveraged different transformer-based pre-trained language models with fine-tuning to solve this problem. In our analysis, we found that MARBERTv2 outperforms in terms of performance, where macro-F1 is 0.08175 and micro-F1 is 0.61116 compared to other language models that we considered. Our method achieved rank 4 in the testing phase of the challenge.