Glossário
Deduplication
Deduplication is the process of recognising that two or more records describe the same entity and deciding what to do with them: merge, cross-reference, or mark as distinct. The hard part is not deleting a duplicate — it is proving that two similar records are in fact the same one.
In practice
What this changes in your master data
Similarity is not identity
Two materials with similar descriptions may be different gauges. A rule that merges on textual similarity alone creates a worse error than the original.
The decision needs a trail
Who merged, on what basis, and when. Without it the merge is irreversible in practice, even if the database allows undoing.
Questions
Frequently asked questions about this term
Duplication is not a typing error: it is a consequence of multiple entry points. While the door stays open, cleaning is recurring — which is why deduplicating without governing registration is painting over damp.
Partly. The system proposes pairs ranked by likelihood; confirming the ambiguous cases is human, because the cost of a wrong merge is high and asymmetric.
They keep pointing at the surviving record, and the absorbed record history stays accessible. Deleting the old one without redirecting breaks reports and audits.
Keep going
Related terms
Golden record · Master data · MDM (Master Data Management)