mlsanityData to 14 Oct 2026

LYNX: Analytical Program Synthesis via Grounding and Closed-Loop Refinement

arXiv 2610.14777 ↗First posted 10 Oct 2026cs.AI cs.DBcited by 0

Not on the map. Why?

Abstract (arXiv)

LLM-powered data analytics aims to translate natural language instructions into executable analytical programs over underlying data. However, existing approaches, whether treating LLMs as intent-to-code translators or as standalone analytical engines, often fail to reliably align domain semantics with complex schemas and heterogeneous metadata, resulting in brittle execution and semantically inconsistent analytical outputs in real-world workloads. We present LYNX, a multi-agent framework that formulates LLM-driven analytics as a schema- and knowledge-constrained analysis problem rather than a single-pass generation task. LYNX organizes analytical synthesis into a closed-loop workflow of planning, generation, verification, and refinement through three role-specialized agents that collaboratively enforce schema grounding, semantic consistency, and execution correctness. To address domain ambiguity and schema misalignment, LYNX integrates Semantic-Structural Distillation and Stage-Aware Allocation (SSD-SAA). A delta-guided iterative revision continuously diagnoses and repairs semantic and execution-level failures, substantially improving synthesis robustness. We evaluate LYNX on two NL-to-SQL benchmarks, BIRD-SQL Mini-Dev and the more challenging Spider 2.0-Lite. Experimental results demonstrate strong performance and efficiency, achieving 10.66% to 59.54% execution accuracy across lightweight and commercial models, while reducing token consumption by 70.53% compared with the prior most efficient method. Our framework displays high generalization over different backbones, from open-source, lightweight models to the latest commercial models. Additional case studies on quantitative finance analytics and complex multi-relational SQL workloads further demonstrate the framework's robustness, scalability, and practical applicability in real-world analytical scenarios.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 46 of its 62 references that are in the map. First posted on arXiv 10 Oct 2026.

Builds on

46 of 62 references.

References · 46

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  5. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  6. 2022Emergent Abilities of Large Language Modelscited by 2,512
  7. 2023Retrieval-Augmented Generation for Large Language Models: A Surveycited by 2,212
  8. 2023Code Llama: Open Foundation Models for Codecited by 2,139
  9. 2023Teaching Large Language Models to Self-Debugcited by 988
  10. 2023Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLscited by 688
  11. 2024Large Language Models: A Surveycited by 525
  12. 2023DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correctioncited by 477
34 more
  1. 2022DS-1000: A Natural and Reliable Benchmark for Data Science Code Generationcited by 452
  2. 2024A Survey on RAG Meeting LLMs: Towards Retrieval-Augmented Large Language Modelscited by 452
  3. 2019RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parserscited by 424
  4. 2023Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluationcited by 385
  5. 2024The Faiss librarycited by 376
  6. 2024Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflowscited by 232
  7. 2023Evaluating Large Language Models: A Comprehensive Surveycited by 202
  8. 2024CHESS: Contextual Harnessing for Efficient SQL Synthesiscited by 169
  9. 2023OpenAgents: An Open Platform for Language Agents in the Wildcited by 137
  10. 2023TableLlama: Towards Open Large Generalist Models for Tablescited by 137
  11. 2024TableLLM: Enabling Tabular Data Manipulation by LLMs in Real Office Usage Scenarioscited by 77
  12. 2024Large Scale Transfer Learning for Tabular Data via Language Modelingcited by 76
  13. 2023Chat2VIS: Generating Data Visualisations via Natural Language using ChatGPT, Codex and GPT-3 Large Language Modelscited by 59
  14. 2023TableGPT: Towards Unifying Tables, Nature Language and Commands into One GPTcited by 59
  15. 2024Understanding LLMs: A Comprehensive Overview from Training to Inferencecited by 59
  16. 2024Spider2-V: How Far Are Multimodal Agents From Automating Data Science and Engineering Workflows?cited by 56
  17. 2022A Survey on Text-to-SQL Parsing: Concepts, Methods, and Future Directionscited by 52
  18. 2024RSL-SQL: Robust Schema Linking in Text-to-SQL Generationcited by 52
  19. 2024Before Generation, Align it! A Novel and Effective Strategy for Mitigating Hallucinations in Text-to-SQL Generationcited by 50
  20. 2023ScienceBenchmark: A Complex Real-World Benchmark for Evaluating Natural Language to SQL Systemscited by 45
  21. 2025ReFoRCE: A Text-to-SQL Agent with Self-Refinement, Consensus Enforcement, and Column Explorationcited by 37
  22. 2023Rethinking Tabular Data Understanding with Large Language Modelscited by 36
  23. 2024Towards Understanding the Characteristics of Code Generation Errors Made by Large Language Modelscited by 30
  24. 2025LinkAlign: Scalable Schema Linking for Real-World Large-Scale Multi-Database Text-to-SQLcited by 29
  25. 2023CRUSH4SQL: Collective Retrieval Using Schema Hallucination For Text2SQLcited by 24
  26. 2025AutoLink: Autonomous Schema Exploration and Expansion for Scalable Schema Linking in Text-to-SQL at Scalecited by 17
  27. 2023Querying Large Language Models with SQLcited by 14
  28. 2024SiriusBI: A Comprehensive LLM-Powered Solution for Data Analytics in Business Intelligencecited by 12
  29. 2023Several categories of Large Language Models (LLMs): A Short Surveycited by 7
  30. 2025Combinatorial Optimization for All: Using LLMs to Aid Non-Experts in Improving Optimization Algorithmscited by 7
  31. 2024SQL-Encoder: Improving NL2SQL In-Context Learning Through a Context-Aware Encodercited by 5
  32. 2025ODIN: A NL2SQL Recommender to Handle Schema Ambiguitycited by 4
  33. 2025LLM and Agent-Driven Data Analysis: A Systematic Approach for Enterprise Applications and System-level Deploymentcited by 4
  34. 2025Knowledge Distillation with Structured Chain-of-Thought for Text-to-SQLcited by 1

Built on by

Not cited yet.