Is it me you're searching for?
summary
Compares quick ILIKE filtering, external engines like ElasticSearch/Sphinx and built-in Postgres full-text search, arguing Postgres often suffices; demonstrates converting inputs to tsquery, documents to tsvector, using the @@ operator plus ts_rank_cd and weighted vectors (A/B/C/D) for relevance, language-aware stemming and ts_headline for snippets, and storing a tsvector column updated on save to avoid recomputing. Shows a lyrics demo at rubyday.herokuapp.com and reports production-ish timings: 57 ms per search / 27 ms to index.
AI-written summary · about
chapters (8)
- 0:00Introduction and diversity
- 4:59Why search is hard
- 9:53Full text search options
- 11:42Postgres full text search
- 16:21Query, vector and rank
- 22:16Advanced Postgres features
- 29:25Live demo and code
- 35:45Production tips and Q&A
related talks
42mPostgreSQL Tips & Tricks For App Devs
36mUnlocking Smarter Typeahead Search
17mSyntactic Code Search with Sequence-to-Tree Matching: Supporting Syntactic Search with(…)
25mBrowser History Confessional: Searching My Recent Searches
10mTo search or not to search: it depends on the question
45mFrom ChatGPT User to RAG Implementer: A Developer's Journey
13mThinking of Topic Modeling as Search
30mData Rainbows - select * from cloud;