arabic-sentiment-analyzer

Arabic Sentiment Analyzer: Integrating State-of-the-Art Pre-trained Sentiment Model AraBERT, and AraNet in a Single Tool. You just need to download the models from below and it will run properly on your local device!

Note: AraBERT was fine-tuned on ASTD-Unbalanced dataset and AraNet model was utilized as the authors provided.

Demo

arabic-sentiment-analyzer-demo.mp4

How to Use

1. Clone the repo and navigate to the project:

   git clone https://github.com/moayadeldin/AraNet
   cd arabic-sentiment-analyzer

2. In order to use AraNet, you may need to install the forked version of the original repo that addressed the dependencies issues and install it.

   git clone https://github.com/moayadeldin/AraNet
   pip install -e .

3. Start using the app

   python -m streamlit run app.py

Download the Models

AraBERT ASTD-Unbalanced

AraNet

References

AraBERT: Transformer-based Model for Arabic Language Understanding (Antoun et al., OSACT 2020)

Abdul-Mageed, M., Zhang, C., Hashemi, A., & Nagoudi, E. M. B. (2020). AraNet: A Deep Learning Toolkit for Arabic Social Media. arXiv.