Snapfood-Sentiment-Analysis

Overview

This project implements a comprehensive sentiment analysis system for Snapfood restaurant reviews using various machine learning approaches, from traditional algorithms to advanced deep learning models. The system is designed to accurately classify customer sentiments from Persian text reviews.

Getting Started

Clone the Repository

git clone https://github.com/AminRezaeeyan/SnappFood-Sentiment-Analysis.git
cd SnappFood-Sentiment-Analysis

Installation Options

Option 1: Using Docker (Recommended)

1. Make sure you have Docker and Docker Compose installed on your system
2. Build and run the container:

docker-compose up --build

The application will be available at http://localhost:8501

Option 2: Manual Installation

1. Create a virtual environment (recommended):

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

2. Install the required packages:

pip install -r requirements.txt

3. Run the application:

streamlit run app/app.py

Technical Details

Traditional Machine Learning Approaches

The project initially explored several traditional machine learning algorithms:

• Logistic Regression: Used as a baseline model for binary classification
• Random Forest: Implemented with optimized hyperparameters for better feature importance analysis
• XGBoost: Utilized for its superior performance in handling imbalanced datasets

Deep Learning Models

The project implemented various neural network architectures:

• RNN (Recurrent Neural Network): Basic implementation for sequence modeling
• LSTM (Long Short-Term Memory): Used to capture long-term dependencies in review texts
• GRU (Gated Recurrent Unit): Implemented as a more efficient alternative to LSTM

Transformer-Based Approach

The final and most successful implementation uses:

• ParsBERT: A pre-trained Persian BERT model
  • Fine-tuned on the Snapfood review dataset
  • Achieved state-of-the-art results in sentiment classification
  • Optimized for Persian language understanding

Demo

Results

The fine-tuned ParsBERT model achieved the best performance in sentiment classification, demonstrating superior accuracy in understanding Persian language nuances and context in restaurant reviews.