How sanaya27/ai-research-paper-summarizer works
Project type and purpose
• This is a modular, Python-based research paper summarization system. It comprises two primary backend components (PDF extractor and summarizer), a future frontend, an integration layer, a datasets store, and documentation. It is not yet a full end-to-end web app but is structured as discrete services that will eventually be orchestrated by a frontend or orchestration script (app.py).
File structure
• Top-level:
• app.py, pdf_utils.py, summarizer.py, test_pdf_extract.py – likely the orchestration & shared utilities
• pdf-text-extractor-backend/ – Khushboo’s extractor microservice
• summarizer/ – summarization microservice using LLM inference
• frontend/ (placeholder) – to be added by Riya
• integrations/ – code for tying services together and calling Gemini API
• datasets/ – storage of raw PDFs & metadata
• docs/ – architecture, methodology, flow docs
• Each microservice has its own main.py, llm_infer.py (or extraction logic), requirements.txt, sample inputs/outputs, and writes extracted_text.txt & summary.txt.
• .devcontainer, GitHub configs (.gitignore), license, README.md for overall orchestration.
README for insights
• Purpose: extract and preprocess PDF text, then summarize with Google Gemini 1.5 Pro.
• Team-driven, modular.
• Tech stack: Python 3.11+, PyMuPDF, dotenv, Google Gemini API, Streamlit (future UI), Streamlit Cloud.
• Deployment: currently local backends; frontend & integration to follow.
Final advice
• Be detailed: represent each folder/module as a component if it encapsulates a distinct responsibility
• Critical config files (.env handling, devcontainer) as annotations or notes
• Don’t over-segment—keep the pipeline flow clear and focused on extraction → summarization → output