CEIR Publication Sync

Sync Workflow

1. Fetch publications from Charité API for all team members with tokens
2. Enrich with PubMed metadata where DOI/abstract missing
3. Deduplicate by DOI (primary) then normalized title (fallback)
4. Filter out pre-affiliation papers (see filtering rules in references)
5. Compare against existing Zotero items (match DOI or title)
6. Create missing items in Zotero, assign to year collections

Charité Forschungsdatenbank API

Primary source. No browser needed — call REST endpoints directly.

BASE = https://forschungsdatenbank.charite.de/experts/expert

GET /publications/pub_per_exp/{token}/FPS   → publications
GET /exp/co_per_exp/{token}/FPS             → co-authors
GET /exp/info_per_exp/{token}               → profile info

Use 0.3s delay between requests. Returns: title, authors, year, DOI, journal, volume/issue/pages, abstract, PubMed/PMC links, open access status, internal co-authors with tokens.

Implementation: src/zotero_comfort/external/charite.py — ChariteClient class, CEIR_TEAM roster.

Other Sources

• PubMed (src/zotero_comfort/external/pubmed.py): Enrich metadata, verify DOIs, get PMIDs
• ORCID (not yet implemented): 5 members have ORCID only — see references/roster.md
• arXiv (src/zotero_comfort/external/arxiv.py): Low priority, most team output is journals/conferences
• BIH website: Just links to Forschungsdatenbank profiles, no publications inline. Skip.

Deduplication

Publications appear under multiple members. Deduplicate with:

1. DOI (case-insensitive, stripped) — strongest
2. Normalized title (lowercase, non-alnum removed, first 50 chars) — fallback

First occurrence wins. See ChariteClient._deduplicate().

Zotero Sync Target

• Library: CEI_Publications (ID 5767153), key at ~/.secret/ceir-os-zotero-token (format: library_id:api_key)
• Collections: Year-based (2018–2026)
• Item mapping: See workflows_external.py → _paper_to_zotero_item()
• Blocker (2026-02-04): API key is read-only. Need write access.

Reference Files

• [references/roster.md](references/roster.md) — Full team roster with tokens, ORCIDs, and coverage gaps. Read when discovering tokens or checking member coverage.
• [references/filtering.md](references/filtering.md) — Inclusion/exclusion rules, known false positives per member, keyword heuristic. Read when deciding which publications to include or exclude.