Academic Paper Summarizer using ScaleDown

A research-oriented tool that summarizes long academic papers using prompt compression to preserve technical accuracy while reducing processing time and token usage.

This project demonstrates how compression-based prompt optimization can be applied to real-world NLP tasks such as academic paper summarization.

---

🚀 Features

• 📄 Summarizes lengthy academic papers
• 🧠 Uses ScaleDown compression for efficiency
• 📉 Displays compression metrics (tokens saved, ratio, latency)
• ⚙️ Simple Python-based CLI workflow
• 🔬 Designed for research and experimentation
• 📚 Topic extraction from academic papers
• 📝 Dual-mode summarization:
  • Quick Revision Mode (3–4 lines per topic)
  • Deep Study Mode (user-defined word limit per topic)

---

🏗️ Project Structure

Academic-Paper-Summarizer/
│
├── app/
│   ├── papersummarizer.py   # Main summarization script
│   └── sample_paper.txt     # Example academic paper
│
├── scaledown/               # ScaleDown compression library
├── README.md
└── .gitignore

---

📦 Installation

Clone the repository:

git clone https://github.com/Tenzai-AJ/Academic-Paper-Summarizer.git
cd Academic-Paper-Summarizer

Create and activate a virtual environment (recommended):

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

Install dependencies:

pip install -r scaledown/requirements.txt

---

▶️ Running the Project

Move into the app directory:

cd app

Run the summarizer:

python papersummarizer.py

---

📊 Example Output

=== COMPRESSED SUMMARY ===
Programming has evolved beyond traditional text-based languages...

=== COMPRESSION METRICS ===
Original Tokens: 118
Compressed Tokens: 106
Compression Ratio: 1.11
Savings Percent: 10.17
Latency Ms: 52
Model: gpt-4o

===TOPICS IDENTIFIED===

1. Evolution of Programming Languages
2. Compiler Design Concepts
3. Future Trends in Software Development

Choose an Option
1->Quick explanation of all topics
2->Detailed explanation of one topic
3->Exit
Enter choice:

---

🧠 How It Works

1. The academic paper is loaded as raw text
2. ScaleDown compresses the input prompt
3. The compressed prompt is sent to the LLM
4. Major topics are extracted from the paper
5. User selects summarization mode:
   • Quick Revision Mode (concise per topic)
   • Deep Study Mode (detailed per topic with word limit)
6. Compression metrics are displayed for transparency

---

✨ Creative Feature

📚 Study-Oriented Design

The summarizer is designed to support two real-world learning scenarios:

• Rapid exam revision using concise topic summaries
• Conceptual deep dives with user-controlled detail

This bridges prompt optimization with practical academic use.

Explainable Compression Mode

When enabled, the tool exposes detailed metrics such as token savings, compression ratio, latency, and model used — helping users understand how and why compression improves performance.

This makes the project both a summarizer and a learning tool for prompt optimization.

---

📌 Use Cases

• Research paper summarization
• Studying prompt compression techniques
• Efficient LLM usage for long documents
• NLP experimentation and demos

---

🔒 Security Notes

• No API keys are hardcoded
• Environment variables are excluded via .gitignore
• Safe to publish publicly

---

📣 Build in Public

This project was built as part of a learning challenge focusing on:

• Practical LLM usage
• Prompt compression
• Research-oriented tooling

---

📝 License

This project is for educational and experimental purposes.