conference-papers-maintain

name	conference-papers-maintain
description	Use when the user asks to maintain an existing conference paper website by adding papers, merging topics, deduplicating records, enriching metadata, fixing related-work links, or rebuilding after updates.

Conference Papers Maintain

Use this thin entrypoint for incremental maintenance. The implementation lives in ../conference-papers; keep all persistent data in the target site's data/ directory.

Workflow

1. Inspect the existing data/papers/*.json and data/conferences/*.json.
   • Use ../conference-papers/config/user-config.json and optional site-local config/user-config.local.json for default paths, conference, year, topics, and sources.
2. Add or merge new papers:

python3 ../conference-papers/scripts/add_papers.py --draft
python3 ../conference-papers/scripts/add_papers.py --conference CVPR --year 2026 --topic VLA --url "SOURCE" --data data --draft

3. Reclassify topics when topic definitions change:

python3 ../conference-papers/scripts/classify_topics.py

4. Enrich metadata from arXiv when conference pages only provide titles or weak metadata:

python3 ../conference-papers/scripts/enrich_arxiv.py

5. Enrich metadata from available HTML:

python3 ../conference-papers/scripts/enrich_fulltext.py

6. Archive external figure URLs when the user wants local image fallback:

python3 ../conference-papers/scripts/archive_figures.py --data data

7. Validate final note structure and figure references:

python3 ../conference-papers/scripts/validate_notes.py --data data

8. Resolve internal links after related/future work changes:

python3 ../conference-papers/scripts/resolve_links.py

9. Rebuild the site:

python3 ../conference-papers/scripts/build_site.py

Maintenance Rules

• Merge duplicate papers by title or ID instead of creating duplicates.
• Preserve existing final notes when adding topics or metadata.
• Use conference-papers-read for final note upgrades; maintenance should not fake full-paper reading.