Show HN: Hybrid search of 540K+ US Government datasets on 2 CPU cores. No LLMs

Category: infrastructure

Tags: government-data, hybrid-search, data-discovery

Score: 6.3/10 (Innovation: 6, Technical: 6, Documentation: 4, Utility: 7)

This project provides a hybrid search engine over 540K+ US government datasets, combining exact term matching with conceptual search on just 2 CPU cores and without LLMs. It addresses a real gap in discoverability of public data by solving issues like typos and synonym mismatch, making it highly useful for researchers and data journalists.

Target audience: data engineers, data scientists, researchers, journalists

Repository: https://findgovdata.org

View on Hacker News