query-builder

Vespa YQL queries use a SQL-like syntax with operators for text matching, grouping, and ranking. You construct queries by combining selection operators with rank profiles to retrieve and order documents by relevance scores.

A Vespa rank profile defines scoring expressions that evaluate document relevance during ranking. Rank profiles support tensor inputs and ML model integration, enabling dynamic ranking calculations based on query-time features and document attributes.

Yes, Vespa YQL supports the nearestNeighbor operator for approximate nearest neighbor search and accepts query tensors as inputs. These integration points allow you to perform vector similarity searches within the YQL query structure.

Vespa YQL supports predicate filters through its operator set and time-based grouping through the grouping syntax. You combine these features within a single query to filter results and aggregate documents by time intervals.

Vespa YQL provides query timing controls that let you manage execution deadlines and timeouts within your search requests. These controls ensure queries return results within specified time constraints for responsive applications.

Yes, Vespa YQL supports the full YQL operator set, including userQuery integration, grouping syntax, and tensor-based inputs. This enables complex search scenarios combining text matching, filtering, and ML-driven dynamic ranking.