Definition
Semantic search — Semantic search matches a query to results based on meaning and intent rather than exact keyword overlap. A traditional keyword search for "React developer" only returns documents containing that literal phrase, missing a candidate whose resume says "front-end engineer using React and TypeScript" if the exact phrase doesn't appear. Semantic search instead represents both the query and the underlying text as embeddings, numerical vectors that capture meaning, and finds matches based on how close those vectors are, regardless of whether the exact words line up. That's what lets someone search in plain, natural language, "candidates with strong debugging skills who've worked on backend systems," and get relevant results even when no candidate's resume uses that precise phrasing.
Semantic search matters most wherever the exact wording of a query and the exact wording of what's being searched are unlikely to match, which describes most real-world recruiting search.
A recruiter's mental model of what they're looking for rarely matches the exact vocabulary a candidate happened to use in their resume or profile, and semantic search closes that gap by matching on underlying meaning instead of requiring both sides to use the same words.
The Intervieux angle
Intervieux's employer talent search runs semantic, natural-language search over pgvector, letting an employer type a plain description of who they're looking for and search across both their own talent pool and the opt-in, consent-gated Intervieux candidate network.
Frequently asked questions
How is semantic search different from keyword search?
Keyword search matches exact words or phrases. Semantic search matches meaning, using vector representations of text, so a query and a result can be relevant to each other even when they don't share the same wording.
Does semantic search require typing a specific search syntax?
No, that's part of the point. Semantic search is built to work with plain, natural-language queries, since it's matching on meaning rather than requiring exact operators or phrasing.
What powers semantic search under the hood?
Embeddings, numerical vector representations of text that capture semantic meaning, are the core technology. A search compares the vector of the query against the vectors of stored content to find the closest matches.
Related pages
Search talent by what a candidate actually offers
Intervieux's semantic talent search lets employers describe who they're looking for in plain language and search across their own pool and the Intervieux network.