RAG Pipeline with Supabase pgvector
This lesson covers building a RAG pipeline with pgvector in Supabase for semantic search.
Like a library that maps books by topic not just title — RAG converts text into numbers representing meaning, finding similar documents even if the words differ.
- Embedding
- A numerical representation of text in vector space that preserves the text's meaning.
- pgvector
- A PostgreSQL extension for storing and searching vectors directly in the database.
- HNSW
- Hierarchical Navigable Small World — an index algorithm for fast nearest-neighbor search.
- Cosine similarity
- A similarity measure between vectors based on the angle between them — 1 = identical, 0 = orthogonal.
