Semantic Search
Pencarian Semantik
Search that matches on meaning rather than on the exact words typed.
Semantic search finds documents by the meaning of a query rather than by matching keywords. The query is embedded, and documents whose embeddings sit nearby are returned, so a search for "cara cancel langganan" can find a document titled "Cara membatalkan langganan Anda".
It handles wording variation far better than keyword search and matches nothing when the words are precise. The usual production answer is both: keyword search for exact identifiers, codes, and names, semantic search for concepts, then rerank the combined list.
The ranking step is where quality is won or lost. Retrieval can bring back twenty plausible passages and still put the best one last, and the model then answers from the wrong passage with total confidence.
Related terms
Embedding
A list of numbers that places text near other text with similar meaning.
Vector Database
Storage and search built for embeddings rather than exact values.
RAG
Fetch the relevant documents first, then let the model answer using them.
Retrieval
The step that finds the material a model will answer from.
Grounding
Tying the answer to something checkable, so it can be verified instead of trusted.
Want this working inside your team?
It starts with one conversation about your workflow.