Hive Graph
Knowledge graph and entity resolution system for processing large datasets of nonprofit organizations and financial filings from public government sources.
Hive Graph is an entity resolution and knowledge graph system built to make sense of massive public datasets — particularly IRS nonprofit filings, corporate registrations, and government contracts.
The core challenge: the same organization appears differently across datasets. "ACME Foundation", "The ACME Foundation Inc.", and "ACME FOUNDATION" are the same entity. Hive Graph applies a combination of deterministic matching rules and probabilistic scoring to resolve these into canonical entities, then links them into a graph.
Built entirely on Cloudflare Workers and D1. The processing pipeline ingests raw CSV/JSON, runs entity extraction and normalization, and writes resolved records into D1. Graph traversal queries expose relationship paths between any two organizations.