Semantic Caching for OLAP via LLM-Based Query Canonicalization (Extended Version)

2026-02-23

AI Summary

This work proposes the first semantic caching mechanism tailored for star-schema OLAP workloads that enables safe reuse of semantically equivalent queries across diverse interfaces—including BI tools, notebooks, and natural language systems—where traditional caches based on SQL text or AST keys fail. The approach leverages large language models to normalize both SQL and natural language queries into a unified “OLAP intent signature,” augmented with rigorous schema validation and confidence gating to guarantee zero false cache hits. To broaden applicability, it introduces two fidelity-preserving inference strategies: roll-up and filter-down. Evaluated on 1,395 queries from TPC-DS, SSB, and NYC TLC benchmarks, the method achieves an 82% cache hit rate—substantially outperforming text-based (28%) and AST-based (56%) baselines—and doubles the hit rate for hierarchical queries.

Technology Category

• Data Mining & Knowledge Management: Intelligent Query Processing
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Search and Optimization: Distributed Search

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Analytical workloads exhibit substantial semantic repetition, yet most production caches key entries by SQL surface form (text or AST), fragmenting reuse across BI tools, notebooks, and NL interfaces. We introduce a safety-first middleware cache for dashboard-style OLAP over star schemas that canonicalizes both SQL and NL into a unified key space -- the OLAP Intent Signature -- capturing measures, grouping levels, filters, and time windows. Reuse requires exact intent matches under strict schema validation and confidence-gated NL acceptance; two correctness-preserving derivations (roll-up, filter-down) extend coverage without approximate matching. Across TPC-DS, SSB, and NYC TLC (1,395 queries), we achieve 82% hit rate versus 28% (text) and 56% (AST) with zero false hits; derivations double hit rate on hierarchical queries.

Problem

• Semantic Caching
• OLAP
• Query Canonicalization
• Cache Reuse
• Natural Language Queries

Innovation

• Semantic Caching
• Query Canonicalization
• OLAP Intent Signature
• LLM-based NL Understanding
• Correctness-preserving Derivation