Duplicates and relations
Detection compares documents as they arrive and proposes pairs. Nothing is merged either way — a person decides.
Auto-merging a near-duplicate is how the one copy carrying a handwritten note in the margin disappears. Detection proposes; a person confirms.
Answering a proposal
A banner on the documents tab counts the pairs awaiting a decision. Opening it shows each pair with a similarity percentage where one could be computed.
Two answers:
- Same document — recorded as the same document.
- Different — recorded as two different documents.
Either way the decision is stored with your name and the time, and nothing is merged or deleted. Answered pairs move to Already decided.
The relations a pair can carry
| Relation | Inverse |
|---|---|
| Duplicate of | Duplicate of |
| Near-duplicate of | Near-duplicate of |
| Supersedes | Superseded by |
| Attachment of | Has attachment |
| Exhibit of | Has exhibit |
| Responds to | Answered by |
| Excerpt of | Excerpted in |
They appear under Related documents on the detail sheet, from both sides.
Versions
A document can carry several versions — v2 — current, v1 — superseded — each recording who filed it. The current version is what the viewer opens and what the analysis reads.
Arrivals versus duplicates
These are different things.
An arrival is the same bytes reaching the firm again — from another custodian, or by another route. Filing them records another arrival on the existing document rather than creating a second one. One stored object, several rows under Where it came from.
A duplicate is two documents whose text is similar enough that they may be the same thing. That is a judgement, and it is yours.