📚 Mini Research Assistant (RAG-based)

A research assistant prototype that helps you quickly find, rank, and summarize scientific papers on a given topic using retrieval-augmented generation (RAG) with OpenAI GPT models.

The system fetches paper abstracts, creates embeddings, performs similarity search (FAISS), reranks results with a cross-encoder, and summarizes relevant papers for your query. Includes a simple Streamlit UI for interactive use.

---

⚡ Features

• Search scientific papers by topic and preview abstracts.

• Semantic ranking using FAISS embeddings.

• cross-encoder reranking for improved relevance.

• Chunking support for both abstracts and full-text papers (PDFs).

• Summarization of top relevant papers with an LLM (OpenAI GPT).

• Streamlit-based interactive UI:

  • Enter a topic
  • Search abstracts
  • View summaries
  • Copy or download summaries

---

🛠 Installation

1. Clone the repository:

git clone https://github.com/spiridonoff/research-assistant.git
cd mini-research-assistant

2. Create a virtual environment (optional but recommended):

python -m venv venv
source venv/bin/activate  # Linux/Mac
venv\Scripts\activate     # Windows

3. Install dependencies:

pip install -r requirements.txt

4. Set your OpenAI API key:

export OPENAI_API_KEY="your_api_key_here"  # Linux/Mac
setx OPENAI_API_KEY "your_api_key_here"    # Windows

---

🚀 Usage

Run the Streamlit app:

./run.sh

or manually:

export PYTHONPATH=src
streamlit run src/app/main.py

Workflow:

1. Enter a topic to fetch related papers.
2. Preview the first few abstracts.
3. Enter a research query to search across abstracts.
4. View ranked results and summaries.
5. Copy or download the summaries for further use.

---

📁 Project Structure

src/
├─ app/
│  └─ main.py        # Streamlit UI
├─ rag/
│  ├─ io/
│  |  ├─ fetch_abs.py
│  │  ├─ fetch_papers.py
│  │  └─ text_utils.py
│  ├─ index/
│  │  ├─ build_index_abs.py
│  │  ├─ build_index_paper.py
│  │  ├─ search_abs.py
│  │  └─ search_paper.py
│  ├─ pipelines/
│  │  └─ summarizer.py
├─ config.py          # API keys and configuration
run.sh                # Launcher script with PYTHONPATH
requirements.txt

---

⚡ Next Steps / Future Improvements

• Add selection & download of specific papers.
• Integrate OLMo or other open source LLMs for research summaries.
• Improve prompt design for better summaries.
• Extend UI for follow-up questions using conversational LLM.

---

💡 Notes

• This project is intended as a mini prototype / learning project.
• Designed to be modular: abstracts search, embedding, FAISS indexing, reranking, summarization, and UI can be extended independently.
• OpenAI API usage may incur costs depending on your queries.

---

📝 License

This project is licensed under the MIT License. MIT License – feel free to reuse and modify.