Skip to content

CLAUDE.md

Latest commit

 

History

History
67 lines (52 loc) · 3.94 KB

CLAUDE.md

File metadata and controls

67 lines (52 loc) · 3.94 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

This is a comprehensive patent analytics toolkit designed for patent information professionals, PATLIB staff, and patent office personnel. The repository contains multiple specialized modules for different aspects of patent data analysis, from interactive web applications to Jupyter-based analytical workflows. The toolkit demonstrates various approaches to patent data analysis using EPO data sources and modern visualization techniques.

Development Guidelines

Important Workflow Tips

  • Always use NotebookEdit when reading or changing Jupyter Notebooks to prevent errors like "Error: File is a Jupyter Notebook. Use the NotebookEdit to edit this file"
  • Notebook Cell Best Practices:
    • Use markdown cells for documentation, explanations, and static information
    • Use code cells only for executable code and dynamic outputs
    • Avoid code cells that only print static text or documentation
    • Keep notebooks clean and focused on functionality, not verbose explanations

EPO OPS API Integration (Key Findings)

Working Configuration for German University Patents:

  • Endpoint: published-data/application/epodoc/EP{number}/biblio
  • Authentication: OAuth2 with credentials in ipc-ops/.env
  • Format Processing: Remove "EP" prefix, remove leading zeros, remove kind codes (A/B)
  • Example: EP19196837A → published-data/application/epodoc/EP19196837/biblio

Critical Discovery: German university patents from DeepTechFinder are application numbers, NOT publication numbers. Must use /application/ endpoint instead of /publication/.

OPS Response Structure for Data Extraction:

  • Applicants: Search for keys containing 'applicant' → extract 'applicant-name' + 'residence.country'
  • Inventors: Search for keys containing 'inventor' → extract 'inventor-name' + 'residence.country'
  • Priority Claims: Search for 'priority-claim' → extract 'country' + 'doc-number' + 'date'
  • Application Info: Search for 'application-reference' → extract 'document-id' components
  • Title: Search for 'invention-title' → extract text content (may be in '$' or '#text' fields)

Verified Working Patents:

  • EP19196837A (Technische Universität Dresden)
  • EP18826058A (University of Applied Sciences Saarbrücken)
  • EP09735811A (University of Applied Sciences Saarbrücken) - required leading zero fix

Critical Leading Zero Discovery:

  • Patents from 2000s (EP09735811A) require leading zero preservation: EP09735811 not EP9735811
  • Older patents (EP80100298A) can have leading zeros removed: EP80100298
  • Implement fallback strategy: try with leading zero first, then without if 404

Rate Limiting: Standard EPO OPS limits apply - implement delays between requests.

Data Extraction Issues Resolved:

  • Duplicate Data: EPO OPS returns multiple formats (epodoc, original) for same entities
  • Solution: Prefer 'original' format for cleaner names, deduplicate entries
  • Priority Claims: Extract from priority-claims section with proper date formatting
  • CPC Classifications: Located in patent-classifications not classifications-cpc
  • IPC Formatting: Clean spaces from "G01B 5/ 00" → "G01B5/00"
  • Title Selection: Prefer English (@lang="en"), fallback to first available

Production Notebook Structure:

  • Comprehensive markdown documentation for patent searchers
  • Professional presentation format with clear methodology
  • Each code cell has purpose statement and searcher context
  • Export capabilities for further analysis (CSV with complete bibliographic data)

For Python/Jupyter Development

  • Use virtual environments for dependency management
  • Follow existing code patterns in each module
  • Test with PATSTAT TEST environment before PROD
  • Document new analysis workflows in notebooks

[... rest of the original file content remains the same ...]