The procedure of detecting and removing or consolidating multiple records within a single collection that represent the same bibliographic item or digital object, preserving provenance and intended distinctions (editions, manifestations) while reducing redundant representations.

Principle

Detection combines stable identifiers, normalized metadata and similarity measures; consolidation requires choosing or constructing a canonical record and recording provenance so that information loss is controlled and reversible where necessary.

Demonstration

Illustrative scenario → A library catalog contains two records for the same edition with slight metadata variations. Recognition → matching ISBN and high similarity in title/author identify a duplicate. Action → holdings and local notes are merged onto a canonical record; original records are retained in an audit trail. Consequence → search results and circulation logic operate on a single authoritative record while provenance allows reconstruction.

Misapplication

Merging records that represent distinct manifestations (different editions, translations, or formats) merely because they share a title or author. The error is conflating redundancy with legitimate variant records and losing bibliographic fidelity.

Consequence

Appropriate deduplication reduces user confusion, storage overhead and indexing noise; overzealous deduplication eliminates meaningful distinctions, breaks citation chains and can remove metadata required for preservation or rights management.

Reversal

In preservation, scholarly, or legal contexts duplicates (or variant records) may be intentionally retained; when variants carry materially different metadata or rights, consolidation is inappropriate and deduplication must be constrained or avoided.

Boundary

Clearly within: removing exact or near‑exact duplicate records for the same edition in a single institutional catalog. Boundary case: records for the same work but different editions or translations where policy determines consolidation. Clearly outside: cross‑dataset entity reconciliation (distinct process oriented to linking rather than internal consolidation).

Semantic Tension

Completeness/Fidelity ↔ Simplicity/Efficiency — removing duplicates simplifies access and storage but may sacrifice bibliographic or preservation fidelity.

Synthesis

Deduplication is an editorial process requiring explicit rules and provenance: technical detection must be paired with policy about when consolidation preserves or destroys essential distinctions.