Course Review Sentiment Analysis API

A powerful Flask-based API that performs sentiment analysis on course reviews. This tool uses a machine learning model trained on 1.4 million reviews to predict sentiment (Positive/Negative), extract meaningful keywords, and provide detailed analytics. It features a sleek, modern frontend for easy testing and robust API endpoints for integration with other applications.

Live Demo: https://course-review-sentiment-api.onrender.com

✨ Key Features

• High-Accuracy Predictions: Utilizes a Decision Tree model trained on a massive dataset, achieving a 98.7% test accuracy.
• Advanced NLP Processing: Employs Natural Language Processing (NLP) for intelligent text parsing, cleaning, and lemmatization.
• Intelligent Keyword Extraction: Identifies the most relevant keywords for each review and provides an aggregated list of top positive and negative keywords.
• In-Depth Analytics: Delivers a comprehensive dashboard showing total reviews, sentiment distribution, and the dominant sentiment.
• Dual-Mode Interface:
  • Interactive UI: A simple, modern frontend for direct text input or file uploads (.csv, .txt).
  • Developer API: Clean, accessible API endpoints for seamless integration into other projects.
• Glassmorphic Design: A stunning dark theme with a grainy texture and orangish-red accents for a modern user experience.

🛠️ Technology Stack

• Backend: Python, Flask
• Machine Learning: Scikit-learn
• NLP: NLTK (Natural Language Toolkit)
• Data Handling: Pandas, NumPy
• Frontend: HTML5, CSS3, JavaScript
• Deployment: Render

🚀 Getting Started

Follow these instructions to get a local copy up and running for development and testing purposes.

Prerequisites

• Python 3.8+
• pip (Python package installer)

Installation

1. Clone the repository:

   git clone [https://github.com/dhruv-deb/course-review-sentiment-API.git](https://github.com/dhruv-deb/course-review-sentiment-API.git)
   cd course-review-sentiment-API

2. Create and activate a virtual environment:

   • Windows:

     python -m venv venv
     .\venv\Scripts\activate

   • macOS / Linux:

     python3 -m venv venv
     source venv/bin/activate

3. Install the required dependencies:

   pip install -r requirements.txt

4. Download NLTK Data: The application is configured to download the necessary NLTK packages on the first run.

5. Run the Flask application:

   python app.py

   The application will be available at http://127.0.0.1:5000.

🔌 API Endpoints

The API provides two main endpoints for programmatic access.

1. Analyze Single Text

• Endpoint: /api/analyze-text
• Method: POST
• Body: Raw JSON

Example Request (cURL):

curl -X POST -H "Content-Type: application/json" \
-d '{"text": "This course was absolutely fantastic!"}' \
[http://127.0.0.1:5000/api/analyze-text](http://127.0.0.1:5000/api/analyze-text)

Example Response:

{
  "text": "This course was absolutely fantastic!",
  "sentiment": "Positive",
  "confidence_label": 1.0,
  "keywords": {
    "positive": ["fantastic", "course"],
    "negative": []
  }
}

2. Analyze a File

• Endpoint: /api/analyze-file
• Method: POST
• Body: multipart/form-data

Example Request (cURL):

curl -X POST -F "file=@/path/to/your/reviews.csv" [http://127.0.0.1:5000/api/analyze-file](http://127.0.0.1:5000/api/analyze-file)

Example Response:

{
  "filename": "reviews.csv",
  "sentiment_results": [
    {
      "text": "Great content and clear explanations.",
      "keywords": ["great", "content", "clear"],
      "label": "Positive"
    },
    {
      "text": "The instructor was boring and hard to follow.",
      "keywords": ["boring", "hard"],
      "label": "Negative"
    }
  ],
  "aggregate_sentiment": {
    "total": 2,
    "count": { "Positive": 1, "Negative": 1 },
    "percent": { "Positive": "50.0%", "Negative": "50.0%" },
    "most_common": "Positive"
  },
  "keywords_analysis": {
    "top_positive_keywords": ["great", "content", "clear"],
    "top_negative_keywords": ["boring", "hard"]
  }
}

📂 Project Structure

course-review-sentiment-API/
├── model/                  # Contains the serialized ML model and vectorizers
│   ├── model_destree.pkl
│   ├── countVectorizer.pkl
│   └── scaler.pkl
├── nltk_data/                  # Contains the nltk_data
├── static/                 # CSS and other static assets
│   └── style.css
├── notebook/                 # Contains the jupyter notebook with for training reference
│   └── Data_Exploration_&_Modelling.ipynb
├── templates/              # HTML templates for the frontend
│   └── index.html
├── app.py                  # Main Flask application file
├── utils.py                # NLP processing and helper functions
├── requirements.txt        # Project dependencies
└── README.md               # This file

✍️ Author

• Dhruv Deb - dhruv-deb