Open research method · artwork identity verification
How do you verify artwork identity across museum records?
When two collection or archival records look similar, what evidence supports treating them as the same physical artwork?
Publication is not authorized. This is a local, reviewable release candidate—not a finding about a specific artwork.
Why this question is difficult
Museum and archival records often describe the same object differently, while unrelated works can share translated titles, subjects, makers, and approximate dimensions. A persuasive resemblance is therefore a research lead, not an identity decision.
The Research Commons treats this as a method question rather than a shortcut to a historical conclusion. Work remains source-first, identity-aware, revision-bound, and open to correction. Every consequential result requires an attributable human reviewer, and publication remains a separate approval decision.
A reproducible method
Construct an object fingerprint before reading the provenance narrative. Record every title form, maker assertion and qualifier, measurements with units and orientation, medium, inscriptions, labels, stamps, frame notes, image features, accession or stock numbers, and cited catalogue entries. Compare each field independently and retain mismatches.
Create one machine-readable packet for the exact revision being reviewed. Preserve source URLs, visible queries, access dates, quotations or field locations, object-identity anchors, contradictions, rejected candidates, alternative hypotheses, gaps, and limitations. Hash the canonical packet so later corrections cannot silently alter the evidence reviewed.
Evidence and responsible automation
Prefer stable identifiers and object-specific documentary bridges over label similarity. A dealer stock card, claim number, catalogue raisonné entry, dated photograph, accession trail, or archival correspondence that joins several independent anchors carries more weight than one shared title. Keep carrier, depicted work, photograph, and digital surrogate distinct.
AI agents may assist with bounded query planning, retrieval after a human trigger, evidence extraction, citation checking, collision warnings, contradiction detection, and packet assembly. They may not establish identity, authenticity, title, wrongdoing, novelty, publication readiness, or legal conclusions. Missing evidence should produce a refusal or next research action, not invented completion.
Challenge the result before using it
Actively search for collisions by the same maker, similarly sized works, alternate translations, copies, variants, and records whose dimensions were rounded or transposed. If a material mismatch remains unexplained, preserve separate candidate identities and ask what primary evidence could resolve them.
A qualified reviewer should reproduce the most material citation, challenge identity, test the alternative hypothesis, identify source dependencies, and record corrections. The reviewer’s decision must name what is supported, disputed, unresolved, or outside scope. A polished narrative, repeated database statement, or high similarity score cannot substitute for this work.
Primary method references
- Linked Art data model — Event-centered cultural-heritage data semantics and multi-valued evidence structures.
- Getty Provenance Index — Official scope and access context for provenance, sales, dealer, and archival indexes.
- CIDOC Conceptual Reference Model — Reference model for events, actors, objects, places, timespans, and evidence-bearing relationships.
Questions and limits
How many matching fields prove two records describe the same artwork?
No fixed match count proves identity. The quality and independence of anchors matter, and consequential joins require accountable human review.
Can Meta Museum or an AI agent publish the resulting conclusion automatically?
No. The workflow can assemble evidence and proposed next actions, but publication, novelty, and consequential conclusions always require attributable human approval.