Amharic Sentiment Analysis using Neural Networks

This project is designed to perform sentiment analysis on Amharic text using a neural network model. The model is built using PyTorch for the machine learning component and Flask for the web interface. The project includes preprocessing of Amharic text, training a sentiment analysis model, and deploying it as a web application.

Table of Contents

• Installation
• Usage
• Model Performance
• API Endpoints
• Dependencies
• Contributing
• License

---

Installation

1. Clone the repository:

   git clone https://github.com/Danitilahun/Amharic_NLP_Sentiment_Analysis_Neural_Networks_LSTM.git

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

---

Usage

1. Run the Flask application:

   flask run --port=5001

2. Access the web interface: Open your browser and go to http://127.0.0.1:5001/.

3. Analyze sentiment: Enter Amharic text in the provided input field and submit to get the sentiment analysis result.

---

Model Performance

The model's performance on the train and test datasets is as follows:

Train Metrics:

• Accuracy: 0.9461
• Precision: 0.9757
• Recall: 0.9185
• F1 Score: 0.9464
• AUC: 0.9781

Test Metrics:

• Accuracy: 0.9219
• Precision: 0.9522
• Recall: 0.8937
• F1 Score: 0.9220
• AUC: 0.9575

Classification Report (Train):

• Class 0.0:
  • Precision: 0.9179
  • Recall: 0.9754
  • F1-Score: 0.9458
  • Support: 26757.0
• Class 1.0:
  • Precision: 0.9756
  • Recall: 0.9188
  • F1-Score: Missing
  • Support: Missing

Classification Report (Test):

• Class 0.0:
  • Precision: 0.8939
  • Recall: 0.9529
  • F1-Score: 0.9218
  • Support: 2983.0
• Class 1.0:
  • Precision: 0.9523
  • Recall: 0.8937
  • F1-Score: Missing
  • Support: Missing

---

API Endpoints

The Flask application provides the following API endpoints:

• GET /: Renders the home page with the sentiment analysis form.
• POST /analyze: Accepts JSON data with a text field and returns the sentiment analysis result.

Example Request:

{
  "text": "ይህ ፕሮጀክት በጣም ጥሩ ነው!"
}

Example Response:

{
  "sentiment": "positive",
  "probability": 95.7
}

---

Dependencies

The project relies on the following Python libraries:

• Flask: For building the web interface.
• gunicorn: For deploying the Flask app.
• torch: For building and training the neural network.
• nltk: For text tokenization.
• numpy: For numerical operations.
• pickle: For loading pre-trained embeddings.

You can install all dependencies by running:

pip install -r requirements.txt

---

Note: This project is for educational purposes and may require further tuning for production use.