AI Agent: Hugging Face Paper Analyzer & Notion Curator

Unlock Automated AI Research Curation with this AI Agent

Overview

This AI Agent acts as your personal research assistant, dedicated to keeping you at the forefront of AI advancements. It diligently monitors Hugging Face for new papers from the previous day, leverages OpenAI's sophisticated language models (specifically GPT-4o as configured) to perform in-depth abstract analysis, and meticulously organizes these insights into your Notion workspace. The agent's core abilities include: automated paper scraping, AI-driven data extraction (core introduction, keywords, results, technical details, classification), and structured data storage in Notion, ensuring you never miss a critical publication.

Key Features & Benefits

• Automated Paper Discovery: Schedules daily checks (weekdays by default) on Hugging Face for newly listed research papers.
• AI-Powered Abstract Analysis: Utilizes OpenAI (GPT-4o by default) to intelligently dissect paper abstracts, extracting:
  • Core Introduction & Significance
  • 2-5 Key Keywords
  • Key Data & Performance Results
  • Technical Methods, Optimizations & Datasets
  • Academic Classification(s)
• Structured Notion Integration: Automatically creates and populates entries in your designated Notion database with detailed paper information and AI-generated insights.
• Duplicate Avoidance: Checks your Notion database to prevent adding papers that have already been processed.
• Time Efficiency: Frees you from manual tracking and summarization, allowing focus on strategic application of research.
• Customizable: Easily modify the schedule, specific OpenAI model, or Notion property mapping within the n8n interface.

• An n8n instance (Cloud or self-hosted).
• OpenAI API Key with access to a suitable model (e.g., GPT-4o, which is pre-configured, or other GPT-3.5/4 models).
• Notion API Key and the ID of the Notion Database you want to populate.
• Your Notion database must be pre-configured with properties to store: URL (URL type), title (Title type), abstract (Text type), scrap-date (Date type), Classification (Text or Multi-select type), Technical_Details (Text type), Data_and_Results (Text type), keywords (Text or Multi-select type), Core Introduction (Text type).

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure the 'Schedule Trigger' node to your desired frequency (default is weekdays at 8 AM).
4. In the 'Request Hugging Face Paper' node, verify the URL. The date parameter dynamically fetches papers from the previous day.
5. In the 'Check Paper URL Existed' (Notion) node, select your Notion API credentials and set the correct 'Database ID' where papers will be checked and later stored.
6. In the 'OpenAI Analysis Abstract' node, select your OpenAI API credentials and choose your preferred model (GPT-4o is the default in the template). Review the system prompt to ensure it aligns with your desired analysis depth.
7. In the 'Store Abstract Notion' node, re-confirm your Notion API credentials and the same 'Database ID'. Critically, map the 'Properties' section to your Notion database columns. The workflow is pre-configured to populate fields like 'URL', 'title', 'abstract' (truncated to 2000 characters, adjustable via .substring(0,2000)), 'scrap-date', and the AI-extracted fields ('Classification', 'Technical_Details', etc.). Ensure these property names and types match your Notion setup.
8. Ensure all credentials for Notion and OpenAI are correctly configured and selected in their respective nodes within n8n.
9. Activate the workflow.