Semantic Search
Search by meaning rather than literal words — finds documents that talk about a concept even if they don't use the exact terms.
Semantic Search
Semantic search finds documents by meaning rather than by literal word match. Where keyword search matches “leave policy” against documents containing those two words, semantic search also finds documents about “annual leave”, “PTO”, “time off”, “vacation entitlement” — different vocabularies for the same concept.
Semantic search works by converting both the query and each document into vector embeddings (high-dimensional numeric representations of meaning), then finding documents whose embeddings are mathematically close to the query's.
Papyrus runs semantic search using AI-generated embeddings stored in a managed vector database. Every uploaded document is embedded; queries are embedded on the fly and matched against the store.
Semantic search shines for: conceptual questions, cross-language search, queries by new users who don't know the right vocabulary. It's combined with full-text search in Papyrus's hybrid ranking — you don't pick one or the other.