Semantic Search

ELI5 — The Vibe Check

Semantic search finds results based on meaning, not just keyword matching. Old search: you type 'cheap flights' and it looks for pages with those exact words. Semantic search: you type 'affordable ways to travel by air' and it understands you mean the same thing. It uses embeddings to understand what you MEAN, not just what you TYPE.

Real Talk

Semantic search uses vector embeddings to retrieve results based on meaning rather than lexical matching. Text is converted to dense vectors via embedding models, stored in vector databases, and retrieved using similarity measures (cosine similarity, dot product). This enables finding relevant content even when queries and documents use different terminology.

Show Me The Code

// Semantic search with vector DB
const query = "how to handle errors in async code";
const embedding = await embed(query);

// Finds docs about try-catch, error boundaries,
// promise rejection — even without those exact words
const results = await vectorDB.search(embedding, { topK: 5 });

When You'll Hear This

"Semantic search found the relevant doc even though the user used completely different terminology." / "We replaced keyword search with semantic search and relevance improved 3x."