每日文献雷达：2026-07-14

今日 slides：research-radar-2026-07-14

每日文献雷达：2026-07-14

今日自动检索并筛选出 5 篇候选论文，通过结构化深度阅读生成以下分析。


flowchart TB
    subgraph TODAY["今日入选 2026-07-14"]
    P1["H-STAR: LLM-driven Hybrid SQL-Text Adaptive Reason..."]
    P2["Evaluating NL2SQL via SQL2NL..."]
    P3["Something's Fishy In The Data Lake: A Critical Re-..."]
    P4["CHASE-SQL: Multi-Path Reasoning and Preference Opt..."]
    P5["Enhancing Temporal Understanding in LLMs for Semi-..."]
    end
    TODAY --> BLOG["博客深度阅读"]
    BLOG --> SLIDES["Marp 幻灯片"]
    style TODAY fill:#f0f4ff,stroke:#4a90d9
    style BLOG fill:#d4edda,stroke:#155724
    style SLIDES fill:#e2d9f3,stroke:#6f42c1

今日入选

H-STAR: LLM-driven Hybrid SQL-Text Adaptive Reasoning on Tables（score: 0.3065）
Evaluating NL2SQL via SQL2NL（score: 0.245）
Something’s Fishy In The Data Lake: A Critical Re-evaluation of Table Union Search Benchmarks（score: 0.23）
CHASE-SQL: Multi-Path Reasoning and Preference Optimized Candidate Selection in Text-to-SQL（score: 0.2225）
Enhancing Temporal Understanding in LLMs for Semi-structured Tables（score: 0.211）

H-STAR: LLM-driven Hybrid SQL-Text Adaptive Reasoning on Tables

作者：Nikhil Abhyankar, Vivek Gupta, Dan Roth, Chandan K. Reddy
入选原因：topic keywords matched; 22 citations; appeared 1x before
来源信息：发表：North American Chapter of the Association for Computational Linguistics | 引用：22 | 链接：https://arxiv.org/abs/2407.05952

Tabular reasoning involves interpreting natural language queries about tabular data, which presents a unique challenge of combining language understanding with structured data analysis.

摘要：Tabular reasoning involves interpreting natural language queries about tabular data, which presents a unique challenge of combining language understanding with structured data analysis. Existing methods employ either textual reasoning, which excels in semantic interpretation but struggles with mathematical operations, or symbolic reasoning, which handles computations well but lacks semantic understanding. This paper introduces a novel algorithm H-STAR that integrates both symbolic and semantic (textual) approaches in a two-stage process to address these limitations. H-STAR employs: (1) step-wise table extraction using `multi-view’ column retrieval followed by row extraction, and (2) adaptive reasoning that adapts reasoning strategies based on question types, utilizing semantic reasoning fo…
方法·三元组：当前环境未配置 LLM API Key，无法生成结构化三元组分析。GitHub Actions CI 中会使用百炼 API 进行深度阅读，采用公众号 storytelling 风格输出。本地可通过设置 LLM_API_KEY、LLM_BASE_URL、LLM_MODEL 环境变量启用。
实验：需确认数据集、指标和 baseline。
风险：离线或工具降级时摘要可能不足，不能替代人工精读。
后续动作：深读方法和实验设计，建议人工使用 /readpaper 精读

Evaluating NL2SQL via SQL2NL

作者：Mohammadtaher Safarzadeh, Afshin Oroojlooyjadid, Dan Roth
入选原因：appeared 2x before
来源信息：发表：Conference on Empirical Methods in Natural Language Processing | 链接：https://arxiv.org/abs/2509.04657

Robust evaluation in the presence of linguistic variation is key to understanding the generalization capabilities of Natural Language to SQL (NL2SQL) models, yet existing benchmarks rarely address this factor in a systematic or controlled manner.

摘要：Robust evaluation in the presence of linguistic variation is key to understanding the generalization capabilities of Natural Language to SQL (NL2SQL) models, yet existing benchmarks rarely address this factor in a systematic or controlled manner. We propose a novel schema-aligned paraphrasing framework that leverages SQL-to-NL (SQL2NL) to automatically generate semantically equivalent, lexically diverse queries while maintaining alignment with the original schema and intent. This enables the first targeted evaluation of NL2SQL robustness to linguistic variation in isolation-distinct from prior work that primarily investigates ambiguity or schema perturbations. Our analysis reveals that state-of-the-art models are far more brittle than standard benchmarks suggest. For example, LLaMa3.3-70B …
方法·三元组：当前环境未配置 LLM API Key，无法生成结构化三元组分析。GitHub Actions CI 中会使用百炼 API 进行深度阅读，采用公众号 storytelling 风格输出。本地可通过设置 LLM_API_KEY、LLM_BASE_URL、LLM_MODEL 环境变量启用。
实验：需确认数据集、指标和 baseline。
风险：离线或工具降级时摘要可能不足，不能替代人工精读。
后续动作：保留为快读候选，后续按主题相关性跟进

Something’s Fishy In The Data Lake: A Critical Re-evaluation of Table Union Search Benchmarks

作者：Allaa Boutaleb, Bernd Amann, Hubert Naacke, Rafael Angarita
入选原因：code signal; 3 citations; appeared 4x before
来源信息：发表：Proceedings of the 4th Table Representation Learning Workshop | 引用：3 | 代码：https://github.com/Allaa-boutaleb/fishy-tus | 链接：https://arxiv.org/abs/2505.21329

Recent table representation learning and data discovery methods tackle table union search (TUS) within data lakes, which involves identifying tables that can be unioned with a given query table to enrich its content.

摘要：Recent table representation learning and data discovery methods tackle table union search (TUS) within data lakes, which involves identifying tables that can be unioned with a given query table to enrich its content. These methods are commonly evaluated using benchmarks that aim to assess semantic understanding in real-world TUS tasks. However, our analysis of prominent TUS benchmarks reveals several limitations that allow simple baselines to perform surprisingly well, often outperforming more sophisticated approaches. This suggests that current benchmark scores are heavily influenced by dataset-specific characteristics and fail to effectively isolate the gains from semantic understanding. To address this, we propose essential criteria for future benchmarks to enable a more realistic and r…
方法·三元组：当前环境未配置 LLM API Key，无法生成结构化三元组分析。GitHub Actions CI 中会使用百炼 API 进行深度阅读，采用公众号 storytelling 风格输出。本地可通过设置 LLM_API_KEY、LLM_BASE_URL、LLM_MODEL 环境变量启用。
实验：需确认数据集、指标和 baseline。
风险：离线或工具降级时摘要可能不足，不能替代人工精读。
后续动作：保留为快读候选，后续按主题相关性跟进

CHASE-SQL: Multi-Path Reasoning and Preference Optimized Candidate Selection in Text-to-SQL

作者：Mohammadreza Pourreza, Hailong Li, Ruoxi Sun, Yeounoh Chung, Shayan Talaei et al.
入选原因：topic keywords matched; 157 citations; appeared 4x before
来源信息：发表：International Conference on Learning Representations | 引用：157 | 链接：https://arxiv.org/abs/2410.01943

In tackling the challenges of large language model (LLM) performance for Text-to-SQL tasks, we introduce CHASE-SQL, a new framework that employs innovative strategies, using test-time compute in multi-agent modeling to improve candidate generation and selection.

摘要：In tackling the challenges of large language model (LLM) performance for Text-to-SQL tasks, we introduce CHASE-SQL, a new framework that employs innovative strategies, using test-time compute in multi-agent modeling to improve candidate generation and selection. CHASE-SQL leverages LLMs’ intrinsic knowledge to generate diverse and high-quality SQL candidates using different LLM generators with: (1) a divide-and-conquer method that decomposes complex queries into manageable sub-queries in a single LLM call; (2) chain-of-thought reasoning based on query execution plans, reflecting the steps a database engine takes during execution; and (3) a unique instance-aware synthetic example generation technique, which offers specific few-shot demonstrations tailored to test questions.To identify the b…
方法·三元组：当前环境未配置 LLM API Key，无法生成结构化三元组分析。GitHub Actions CI 中会使用百炼 API 进行深度阅读，采用公众号 storytelling 风格输出。本地可通过设置 LLM_API_KEY、LLM_BASE_URL、LLM_MODEL 环境变量启用。
实验：需确认数据集、指标和 baseline。
风险：离线或工具降级时摘要可能不足，不能替代人工精读。
后续动作：保留为快读候选，后续按主题相关性跟进

Enhancing Temporal Understanding in LLMs for Semi-structured Tables

作者：Irwin Deng, Kushagra Dixit, Vivek Gupta, Dan Roth
入选原因：10 citations; appeared 2x before
来源信息：发表：North American Chapter of the Association for Computational Linguistics | 引用：10 | 链接：https://arxiv.org/abs/2407.16030

Temporal reasoning over tabular data presents substantial challenges for large language models (LLMs), as evidenced by recent research.

摘要：Temporal reasoning over tabular data presents substantial challenges for large language models (LLMs), as evidenced by recent research. In this study, we conduct a comprehensive analysis of temporal datasets to pinpoint the specific limitations of LLMs. Our investigation leads to enhancements in TempTabQA, a dataset specifically designed for tabular temporal question answering. We provide critical insights for improving LLM performance in temporal reasoning tasks with tabular data. Furthermore, we introduce a novel approach, C.L.E.A.R to strengthen LLM capabilities in this domain. Our findings demonstrate that our method significantly improves evidence-based reasoning across various models. Additionally, our experimental results reveal that indirect supervision with auxiliary data substant…
方法·三元组：当前环境未配置 LLM API Key，无法生成结构化三元组分析。GitHub Actions CI 中会使用百炼 API 进行深度阅读，采用公众号 storytelling 风格输出。本地可通过设置 LLM_API_KEY、LLM_BASE_URL、LLM_MODEL 环境变量启用。
实验：需确认数据集、指标和 baseline。
风险：离线或工具降级时摘要可能不足，不能替代人工精读。
后续动作：保留为快读候选，后续按主题相关性跟进

检索说明

检索层：arXiv（deepxiv）+ Semantic Scholar + Google Scholar，每日查询轮换，保证论文多样性。
阅读层：优先获取 arXiv HTML 全文，使用结构化三元组 + 公众号 storytelling 风格拆解论文逻辑。
深度阅读方法论参考 /readpaper 技能。
自动分析用于雷达筛选，重要论文仍需人工复核。

Research Radar

#literature #research-agent #data-analysis-agent

每日文献雷达：2026-07-14

http://zkkk123.cn/2026/07/14/research-radar/2026-07-14-daily-research-radar/

Author

Ke Zhang

Posted on

July 14, 2026

Licensed under

Hello World Next