Tag research documents by role without inventing certainty
Tag a document along separate dimensions: what it is, what it concerns, and what you have checked. Do not mix those into a single label such as “trusted project evidence.” A small, defined taxonomy is easier to maintain and more useful for retrieval than a large collection of personal synonyms.
W3C provenance supplies background for describing objects and activities. Library of Congress format resources distinguish content and format considerations. The tag dictionary below is an original archive convention, not an official investment-research classification.
Separate three dimensions
Document role describes the source object: annual filing, event notice, agreement, correction, presentation, or secondary report. Topic describes the research subject, such as a named project or entity relationship. Inspection status describes your action, such as collected, identity checked, or relevant passage inspected.
Keep format in another field if it helps: PDF and HTML are not research topics. Keep issuer and period as structured identity fields rather than dozens of improvised tags. This reduces the chance that two researchers use different labels for the same basic attribute.
Hypothetical tag dictionary
Assume fictional Moorland Research is a local personal archive, not a real organization or product. Its example dictionary might be:
| Dimension | Allowed example | Definition |
|---|---|---|
| Role | agreement | Document containing contractual text |
| Role | correction | Explicitly modifies or clarifies identified earlier disclosure |
| Topic | project-cove | Named project identity, not generic “growth” |
| Status | identity-checked | Issuer and document identity compared |
| Status | passage-inspected | Named passage and context read |
| Relationship | translation-of | Connect to original source ID |
The dictionary should explain what a tag does not establish. “Identity checked” says nothing about whether every assertion is true. “Agreement” does not prove execution. “Correction” should not be applied to every later update.
Allow multiple roles without losing precision
One filing can contain several event topics or attach an agreement. Tag the filing’s role and link the attachment as its own source object when needed. Do not force an entire document into one event merely because the archive prefers a single label.
Create an exception queue for ambiguous items. If you cannot classify a file after a brief inspection, record the ambiguity and the next action. Avoid inventing a new permanent tag every time a document is unfamiliar. Review recurring exceptions and add a defined category only when it solves a real retrieval problem.
Verify the taxonomy with retrieval tasks
Try to find all inspected agreements about one project, all corrections linked to an original source, and all unread annual reports for a defined entity. If the tags cannot answer those questions, identify which dimension is missing. If the result contains unrelated items, tighten definitions or separate topic from role.
Check consistency on a small sample across contributors or across your own older notes. Disagreement can reveal an unclear definition rather than careless work. Update the dictionary and record any bulk retagging so historical labels remain interpretable.
Limits and common mistakes
Tags are navigation tools, not proof, rankings, or a substitute for reading. A well-tagged document can still be the wrong version or contain an unverified assertion. Keep evidence checks at the source and claim level.
Common mistakes include using “important” as a catch-all, mixing formats with topics, creating near-duplicate labels, and assigning trust through color. Another is changing label meaning without updating older items. Define a few useful dimensions, test retrieval, and expand only when a real task requires it.
A final record check
Define a rule for taxonomy changes before bulk editing. Keep the old label, new label, reason, and affected scope in a short migration note. Otherwise, a search saved months ago may return a different set without explanation. Treat category maintenance as archive work, while keeping it distinct from new source inspection or a stronger research conclusion.
FAQ
Should every document have many tags?
No. Use the minimum fields needed for retrieval and workflow. More labels increase maintenance work and can reduce consistency.
Can a document have more than one topic?
Yes. Preserve separate topics and source locators where needed. Multiple topics do not mean the whole document supports every claim attached to them.