Hacker News Comments - Hybrid Search
Keyword Search + Semantic Search + Filtering + Faceting
This is a collection of 324K HN comments during the month of Oct 2022.
The data is indexed in Typesense using the new
built-in embedding generation
When a search query is typed in "Semantic Search" mode, Typesense automatically generates embeddings for the query using S-BERT, then does a nearest-neighbor search and returns the closest matches.
When a search query is typed in "Hybrid" mode, Typesense does both a regular keyword search and semantic search and combines the results using Rank Fusion.