The core idea
Local records often mix facts, interpretations, and operational state. Separate them so a topic label does not overwrite the original data or a review decision masquerade as a source fact. Preserve units, time context, and identifier rules when they are relevant to the collection.
A workable starting pattern
Write a small data dictionary with examples and counterexamples. Define how records are linked, revised, and retired. Validate required fields, then inspect whether the values actually mean what the schema says. Keep local extensions documented instead of implying that every consumer will understand them automatically.
A collection of publishing experiments might store the source article, test configuration, measured observations, and editorial interpretation separately. A topic tag connects related experiments without replacing those underlying fields.
What this page does—and does not—offer
TagsFeed does not accept uploads, transform visitor datasets, or host custom data. This page provides design guidance for a system you control.
