Sentiment Analysis for Arabic Tweets using Transformers

Built a sentiment classifier for Arabic tweets by preprocessing (normalization, tokenization, diacritics removal) and fine-tuning a BERT-based model. Outperformed traditional ML models (SVM, Logistic Regression) in accuracy and F1-score.

Tech: Python, Transformers, Hugging Face, Scikit-learn

Impact: Enables businesses to monitor customer feedback, brand reputation, and political sentiment in real time.