How do you evaluate conflicting provenance sources?

Why this question is difficult

Conflicting dates, owners, titles, and restitution statements are common because records were created for different purposes and frequently copy earlier publications. Choosing the most authoritative-looking interface can hide dependency between sources and turn repetition into false corroboration.

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

Split each narrative into atomic assertions: object identity, actor role, event type, date, place, source wording, and certainty. Compare assertions rather than paragraphs. Record whether a source observed the event, reported it later, inferred it from another record, or reproduced a database field without independent evidence.

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

Assess proximity, specificity, custody, and independence. Contemporary correspondence may establish intent without proving completion; a later official database may report completion while relying on an unavailable file. Preserve exact quotations, page or record locations, retrieval dates, and links so another researcher can reproduce the disagreement.

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

Generate alternatives that explain the conflict without assuming error or wrongdoing: two similar objects, planned versus completed events, ownership versus custody, old versus revised attribution, calendar or transcription errors, or two sources sharing one upstream record. A human reviewer should decide whether the claim changes, remains disputed, or needs a narrower boundary.

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