home

Who is named in which document — the cross-reference layer

The join between the archive's text and the structured record: which documents name a given organization, nonprofit or person, resolved against the warehouse's canonical entities. Every mention is scored, not just counted — multi-word names only, a penalty when a name is shared, a heavy demotion when a phrase turns up across so much of the corpus that it is language rather than an entity, and a bonus when the warehouse already ties that entity to money, contracts or court records. The reasons column shows the working for each score. Treat a mention as a lead to verify in the document, never as a finding.

Data source: USVI Public Records + USVI Data Warehouse (VI Update)

Tables

entities

entity_id, kind, canonical_name, phrase, n_documents, confidence, specificity, context_json

Many rows

mentions

entity_id, kind, canonical_name, lf_id, title, collection, doc_date, n_hits, confidence, specificity, reasons, records_url

Many rows

Powered by Datasette · Queries took 2174.85ms · Data source: USVI Public Records + USVI Data Warehouse (VI Update)