Detecting Propaganda Techniques in Arabic Tweets

Propaganda is not a modern term, but the fact that we live now in the information era made it more popular than it was before. Social media platforms, which provide a fertile ground not just for sharing information but also for circulating it among various parties regardless of its accuracy, are one of the critical factors that enhance the existence of propaganda. Accordingly, we are exposed to different forms of information: true without any hostile intentions, true but harmful, unintentionally misleading, and intentionally false. Therefore, propaganda is now ubiquitous, and we must be cautious when dealing with content. This project aims to create an annotated dataset that may later help detect propaganda for the Arabic language. It annotated 600 tweets on six crucial Middle Eastern topics: the 2014 and 2021 Gaza wars, the Turkish coup d'état, the killing of Shireen Abu Akleh, the Lebanese-Israeli maritime border agreement, and Arab-Israeli Normalization. While most previous studies focused on developing methodologies for detecting propaganda techniques without providing justifications, this project labelled 600 tweets with one or more propaganda techniques, along with providing reasons behind their use and some examples. The study found that loaded language, name-calling, and smears were the most commonly used propaganda techniques, while flag-waving, black-and-white fallacy, bandwagon, obfuscation, red herring, appeal to authority, reductio ad hitlerum were less frequent.