Project: datalens
81 entity types
Matrix/Architecture/Query Enhancer
DesignDecisionArchitecture

Query Enhancer

Phase B Intelligent Retrieval implements the Query Enhancer for entity extraction and relevant table identification for queries.

Attributes
labelsDesignDecision,Entity
rationaleThe Query Enhancer entity is designed to improve the accuracy and relevance of SQL queries generated from natural language questions. By extracting entities and understanding relationships within the data schema, it enhances the query planning process, making it more precise especially in complex, multi-table scenarios. This leads to better performance and more reliable answers, addressing the challenges faced with large schemas and context window limitations.
alternatives consideredAlternatives considered include: relying solely on keyword heuristics, superficial schema filtering, and full schema parsing without entity extraction. However, these approaches either lack precision or are too prone to errors with large schemas.
decided byThe decision was made by the DataLens architecture team after evaluating the latest research and internal testing.
decision date2026-03-04
tier2
Relationships1 connections
Loading graph...