Hollywood
Entertainment has no Bloomberg — just IMDb, WGA, and spreadsheets. Hollywood resolves them.
InfrastructureRepository
The Problem
Every question about who represents whom, what stage a project is at, or who's worked with whom gets answered by manually cross-referencing IMDb, WGA, TMDB, and trade press. Assistants build the same spreadsheet by hand, every project, and it's stale before the next one starts. Hollywood pulls those sources into one entity graph instead.
The Solution
A local-first entity graph — a Hono API that ingests trade press, WGA credits, TMDB, and Wikidata into one queryable SQLite database of people, titles, companies, and credits.
What it does
- Ingest trade press, WGA credits, TMDB, and Wikidata through dedicated source adapters
- Archive every raw payload by content hash before normalizing it into the graph
- Connect people, titles, companies, and credits in one queryable database
- Track ingest runs, timing, and record counts for every source
What took work
- Resolving the same real-world person across sources that each assign their own ID
- Normalizing four different source formats — RSS feeds, a directory crawl, a REST API, and SPARQL — into one schema
- Preserving raw payloads and provenance so extraction quality can improve after the fact