# Methodology

The dataset is built from the site-owned `stars/_meta.json` classification and each corresponding star record. It exposes only stable index fields: identity, tier, category, element, Four Transformation associations, canonical URL, source reference, review date, and version.

Per-star source references use the recorded classical citation when a page has one; otherwise they use the dataset-wide classification source. The release does not assert that one school is the sole authoritative reading. Where traditions differ, readers should consult the linked canonical page and original texts.

Release rules: factual corrections increment patch; compatible fields increment minor; breaking schema changes increment major.