📄 GistTerms - Terms of Service Document Classifier

🔍 Overview

GistTerms is an advanced Natural Language Processing (NLP) tool that analyzes Terms of Service (ToS) documents and classifies their contents into meaningful risk categories:

• 🔴 Risk: Potentially harmful or unfavorable to the user
• 🟢 Safe: User-beneficial and protective statements
• 🔵 Information: Neutral or purely informative content

💡 The tool also assigns a grade (A–F) to the document based on its sentence classifications, making it easier for users to quickly assess service agreements.

---

✨ Features

• 🚩 Risk Categorization – Tags each sentence as Risk, Safe, or Information
• 🧮 Document Grading – Assigns an A–F score based on the ratio of risky to safe content
• 🧠 Confidence Filtering – Ignores low-confidence predictions to improve accuracy
• ♻️ Deduplication – Removes repetitive sentences for clean analysis

---

🛠️ Tech Stack

• Language: Python
• Libraries:
  • NLP: nltk, transformers
  • ML: PyTorch
• Dataset: adeeteya/termsofservice
• Hardware: CUDA support for GPU acceleration

---

🚀 Installation

1. Clone the repo

   git clone https://github.com/sumedhkumar96/GistTerms.git
   cd GistTerms

2. Create virtual environment

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

3. Install dependencies

   pip install -r requirements.txt

4. Download NLTK tokenizer

   python -m nltk.downloader punkt

---

🧪 Usage

1. Add your ToS document

   • Save your .txt file as input.txt in the project directory

2. Run the classifier

   python final_program.py

3. Get results in terminal
   Example output:

   Classified Sentences:
   
   🔴 Risk:
     - "The service reserves the right to terminate accounts without notice." (Confidence: 0.85)
   
   🟢 Safe:
     - "Users can request data deletion at any time." (Confidence: 0.90)
   
   🔵 Information:
     - "The service uses cookies to improve functionality." (Confidence: 0.75)
   
   🏅 Service Grade: B
   

---

⚙️ How It Works

1. Sentence Tokenization – Splits input using NLTK
2. Preprocessing – Tokenizes and formats sentences for model input
3. Classification – Uses fine-tuned BERT to categorize content
4. Grading – Computes document grade based on sentence types

---

🛤️ Roadmap

• 🌐 Build a web UI for document upload and visual results
• 📄 Add support for PDF and DOCX formats
• 🤖 Improve accuracy via larger datasets
• 📊 Integrate feedback system for live improvements

---

🤝 Contributing

We welcome contributions!

1. Fork the repo
2. Create a new branch

   git checkout -b feature-name

3. Commit your changes

   git commit -m "Add new feature"

4. Push and open a PR

   git push origin feature-name

---