CADENZA：将自然语言意图编译为用于语义查询处理的特定任务算子DAG

Jul 16, 2026

Upload by Eliana

Semantic query processing engines extend relational query execution with semantic operators implemented through model inference over unstructured data. Their optimization is inherently multi-objective: inference dominates latency and monetary cost, while outputs are stochastic and backend-dependent, requiring quality to be optimized alongside efficiency. Existing optimizers cannot treat per-operator intermediate task outputs as relational optimization objects. CADENZA compiles each intent-bound operator instance into typed task DAGs and selects plans under user-specified trade-offs.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages27

Reading Time1 h 8 min

Abstract

Introduction

Semantic query processing engines and semantic operators

Multi-objective optimization challenges

Limits of existing SQPE optimizers

CADENZA overview (from abstract)

CADENZA: Compiling Natural-Language Intent into Task-Specific Operator DAGs for Semantic Query Processing Jaehyun Ha [jhha@dblab.postech.ac.kr](jhha@dblab.postech.ac.kr)[ ](jhha@dblab.postech.ac.kr)GSAI, POSTECH Pohang, Korea Yongjoo Park [yongjoo@illinois.edu](yongjoo@illinois.edu)[ ](yongjoo@illinois.edu)Univ. of Illinois Urbana-Champaign Urbana, USA Wook-Shin Han∗ [wshan@dblab.postech.ac.kr](wshan@dblab.postech.ac.kr)[ ](wshan@dblab.postech.ac.kr)GSAI, POSTECH Pohang, Korea arXiv :2606 .29151v3 [ cs .DB] 13 Jul 2026 Abstract Semantic query processing engines (SQPEs) extend relational query processing with semantic operators that are executed via model inference over unstructured data. Optimizing such queries is inherently multi-objective: model inference dominates latency and monetary cost, and outputs are stochastic and backend-dependent, so quality must be optimized alongside efficiency. Existing SQPE optimizers do not expose each semantic operator instance’s intermediate task outputs as a relational optimization object, leaving optimization unable to filter, reorder, route, threshold, or jointly tune them. We present CADENZA, which compiles each semantic operator instance—a template bound to a natural-language intent—into an intent-specific plan space of typed task DAGs and selects an executable plan under user-specified quality–latency–cost trade-offs. CADENZA introduces task-extended relational algebra (TxRA), a conservative extension of relational algebra with task-specific operators. The logical planner synthesizes seed TxRA plans, applies structural rewrites whose safety conditions are checked from operator dependencies, and enumerates semantics-guided alternatives from alternative-generation templates. The physical planner compiles each task-specific operator into a router over heterogeneous backends and jointly tunes routing cutpoints, backend parameters, and relational thresholds with Bayesian optimization. On SemBench, CADENZA improves the scenario-level averages of quality, latency, and cost by up to +0.49, 165.7×, and 310.3×, respectively, relative to state-of-the-art. Keywords Semantic Operators, Large Language Models, Query Optimization, Multi-Objective Optimization, Task-Specific Operators ACM Reference Format: Jaehyun Ha, Yongjoo Park, and Wook-Shin Han. 2018. CADENZA: Compiling Natural-Language Intent into Task-Specific Operator DAGs for Semantic Query Processing. In Proceedings of (SIGMOD’27) . ACM, New York, NY, USA, 27 pages. [https://doi.org/XXXXXXX.XXXXXXX](https://doi.org/XXXXXXX.XXXXXXX) ∗ Corresponding author. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission [and/or a fee. Request permissions from permissions@acm.org](and/or a fee. Request permissions from permissions@acm.org). SIGMOD’27, Woodstock, NY © 2018 Copyright held by the owner/author(s) . Publication rights licensed to ACM. ACM ISBN 978-1-4503-XXXX-X/2018/06 [https://doi.org/XXXXXXX.XXXXXXX](https://doi.org/XXXXXXX.XXXXXXX) SELECT b.BrandName, COUNT(*) AS NumFlagged FROM Products p JOIN Brands b ON p.BrandId = b.BrandId WHERE p .Category IN ( ' Supplements ' , ' Cosmetics ' ) AND SemFilter ( ' Compliance-sensitive claims (e .g . , "FDA approved") appear on the package labels in {p .Image} . ' ) GROUP BY b.BrandName; Figure 1: A semantic operator instance embedded in SQL. 1 Introduction With the advent of Large Language Models (LLMs), a new class of semantic query processing engines (SQPEs) has emerged [1, 7, 19, 21, 24,

FAQs

What problem does CADENZA address in semantic query processing?

CADENZA addresses the difficulty of optimizing semantic query processing engines where expensive, uncertain model inference makes the task inherently multi-objective (quality, latency, and cost). Existing optimizers lack a way to represent intermediate semantic operator task outputs as relational optimization objects.

How does CADENZA structure optimization of semantic operators?

CADENZA compiles each semantic operator instance—created by binding an operator template to a natural-language intent—into an intent-specific plan space of typed task DAGs, then selects an executable plan that satisfies user-defined trade-offs.

What techniques does CADENZA use to turn logical plans into executable execution?

The logical planner synthesizes seed TxRA plans and applies dependency-safe structural rewrites, enumerating alternatives via semantics-guided templates. The physical planner compiles task-specific operators into routers across heterogeneous backends and jointly tunes routing cutpoints, backend parameters, and relational thresholds using Bayesian optimization.

Jul 16, 2026

Upload by Eliana