Corpus

A corpus, in this platform's usage, is the full body of structured records, court filings, dispositions, enforcement actions, lien data, that its outcome and duration models are built from, assembled deliberately from verified sources rather than accumulated passively. A corpus's real value to an institution pricing risk against it is not its raw ingested record count, which can be inflated by source overlap, temporal re-harvesting, and unverified merged datasets, but its verified unique record count, the figure that survives after every known source of duplication has been checked and removed. Every downstream claim built on a corpus, a base rate, a model's training population, a coverage snapshot, inherits whatever duplication and provenance gaps the corpus contains, which is why this platform treats corpus construction, deduplication, source verification, and provenance labeling, as a standing operational discipline rather than a one-time build completed and then left alone. This platform's own corpus figures are stated with explicit deduplication and verification status attached, distinguishing verified totals from portions still pending recalculation, rather than folding an unverified portion into a headline figure and letting the distinction go unstated, which is the specific practice this platform's own stats-audit discipline exists to catch before an unverified figure reaches a public or acquirer-facing surface.