How engines work
Vector search
Vector search finds documents by comparing embeddings rather than matching keywords, retrieving passages whose meaning is close to the question. It is the retrieval method behind most answer engines.
In more detail
It rewards passages that answer a question completely in one place, because a self-contained passage is what scores well against the question's own embedding.
Guides that cover this
Related terms
Embedding
An embedding is a numeric representation of a piece of text that places similar meanings close together. Embeddings let a system match a question to relevant passages even when they share no words.
Semantic search
Semantic search retrieves results by meaning rather than by exact wording, so a page can be returned for a question that shares none of its words. It underpins how answer engines find material to answer from.
Reranking
Reranking is a second pass that reorders retrieved documents by how well they actually answer the question, before the model writes from them. It decides which of the candidates becomes a citation.
Ready to Dominate AI Search?
Get started now
Allswap