Arabic Sentiment Analysis using RNN, LSTM, and GRU

📌 Project Overview

This project performs sentiment analysis on 100K Arabic reviews using deep learning models:

• Simple RNN

• LSTM

• GRU

🧹 Text Preprocessing

• Arabic text cleaning

• Stopword removal (with negation preservation)

• Tokenization

• Padding

🧠 Models

• Embedding Layer

• Bidirectional LSTM

• GRU

• SimpleRNN

• EarlyStopping

📊 Results

Validation Accuracy Comparison

Confusion Matrix (LSTM)

📦 Dataset

Arabic 100K Reviews Dataset

🚀 Technologies Used

• TensorFlow / Keras

• NLTK

• Scikit-learn

• FastText

👤 Author

Mohamed Mahmoud