Sentiment Analysis of Tweets Related to the Gaza War

Abstract

This research aims to analyze the sentiment of Arabic tweets related to the Gaza conflict. The dataset was built by collecting tweets from social media platforms and previous studies, ensuring a diversity of dialects and perspectives. The tweets underwent preprocessing using a custom Python code, which included removing emojis, standardizing spellings, and removing ineffective common words. Sentiments were then manually classified as positive, negative, or neutral. A modified AraBERT model was trained to accurately perform the classification task. The final system allows for the input of new Arabic tweets and their sentiment analysis with high accuracy.