ziwenhahaha/daily-paper-reader

What it solves

Daily Paper Reader provides a centralized, automated system for researchers to discover and read AI papers. It eliminates the need to manually browse multiple sources by creating a personalized feed of new research from arXiv, OpenReview, and major AI conferences (like NeurIPS, ICLR, ICML, CVPR, etc.), while integrating reading and analysis tools in one place.

How it works

The project uses a "zero-server" architecture leveraging GitHub Actions for automation and GitHub Pages for hosting. It follows a dual-link workflow:

1. Discovery: It automatically fetches new papers from sources like arXiv and OpenReview based on user-defined keywords, research directions, and intent queries. It uses embedding models (such as BGE) and rerankers (like Qwen-Reranker) to filter and prioritize the most relevant papers.
2. Reading & Analysis: Once a paper is selected, the system generates AI-powered summaries and allows users to interact with the paper via an AI chat interface for Q&A. It also extracts figures and tables from PDFs to enhance the reading experience.

Who it’s for

• Individual Researchers: Those who want a personalized "radar" to track specific AI research topics without manual searching.
• Academic Labs: Teams looking to create a shared public paper board to track collective research interests and reading results. -Students: Anyone needing a streamlined workflow to move from discovery to summary and deep reading.

Highlights

• Zero-Server Deployment: Runs entirely on GitHub Actions and GitHub Pages, requiring no paid hosting.
• Personalized AI Feed: Uses semantic search and reranking to tailor paper recommendations to the user's interests.
• Integrated AI Reading: Combines PDF viewing, AI-generated summaries, and a chat interface for real-time paper interrogation.
• Multi-Source Integration: Syncs data from arXiv, OpenReview, bioRxiv, medRxiv, ChemRxiv, and major AI conferences.
• Automated Figure Extraction: Automatically detects and crops figures from PDFs for easier browsing.