About

This is the encyclopedia of the observation theory program, the companion to the textbook Data Mining as Observation. It has one entry for every headword of the book's glossary and three that were filled by hand before the generator existed.

Where the entries come from. An entry is built from the program's records rather than written beside them. The claims ledger in geometric-observation is the authority for what a result is worth, a campaign track file for what was measured, and the book's sources tables for every number the book states. The generator gathers what those records say about one thing, keyed by their identifiers, so that a correction in a record propagates on the next rebuild. The curated parts, the definition, the conditions, the prior art, and the related entries, live in entries.toml.

The card. Every entry opens with a card: the canonical definition and aliases, the exact book version and commit, the epistemic and correction status, the defining equation, the assumptions and scope, the prior-art relationship, the evidence links, and the last semantic review date. Below the card come a schematic, the full entry, the related entries, and what was cited beside the entry without naming it.

Typed edges. A record joins an entry when it names the entry. A book equation defines the entry when the paragraph around it names the entry's terms. A ledger row proves, measures, or refutes and corrects the entry by its class, proved, demonstrated, replicated, predicted, exploratory, refuted, missed, or void. A sources-table row measures the entry when its claim names it. Records cited beside an entry without naming it are listed under see also. A ledger row is printed once in full on the ledger page, and an entry carries its first sentence, its class, and a link to the line.

How to cite. Cite ahb-sjsu/observation-theory-campaigns at the commit on the provenance page, the entry by its id, and for a number the source row the entry gives.