Skip to main content

Classification

When content enters a project with Long-Term Memory enabled, ORQO doesn't just store it — it classifies it. Each piece of content is typed according to the Web-Didaktik ontology and connected to the rest of your knowledge through typed semantic relations. The result is a navigable knowledge graph rather than a flat pile of text.

What Classification Produces​

Two things happen to every piece of content:

  1. It gets a knowledge type. Each unit is classified by what kind of knowledge it is — Orientation, Explanation, Action, or Reference, refined down to a specific type such as a definition, a causal explanation, a procedure, or a reference. This is the function the content plays in understanding, independent of its topic.

  2. It gets connected. Units are linked to each other through typed relations — causal, hierarchical, temporal, foundational, and more. These edges are what let an agent follow a chain of reasoning rather than just retrieve the most similar text.

This is the difference that matters: a similarity index can tell you what looks like a piece of text. A typed knowledge graph can tell you what causes it, what it's a part of, what comes before it in time, and what it's the basis for — connections that often bridge topics sharing almost no words. See Knowledge Navigation for why that changes what agents can do.

When Classification Runs​

Classification happens automatically when:

  • A document is uploaded to a project with Long-Term Memory enabled.
  • A workflow result is captured as knowledge-worthy.
  • An agent explicitly stores a finding.
  • You click Classify on a document in the document library.

It runs asynchronously in the background and reports progress in real time on the project page, so you never wait for it.

_archive Folders Are Excluded​

Documents that live in a folder named _archive (case-insensitive — _Archive, _ARCHIVE, and any nested path like research/_archive/old/ all count) are skipped everywhere: uploads, document-source syncs, the Classify button, the Long-Term Memory backfill, and agent-created documents all land with status skipped and never enter the graph. Container files (PDF, DOCX, etc.) parked in _archive are also not extracted — they stay as the original binary.

Moving an already-classified document into _archive automatically purges its knowledge units from the graph, the same way deleting the document would. Move it back out and it stays skipped — click Classify to re-index it, or toggle Long-Term Memory off and on to backfill the whole library.