mlsanityData to 7 Oct 2026

Semantically Similar, Logically Distinct: Diagnosing the Semantic-Answerability Gap in Table RAG

arXiv 2607.17742 ↗First posted 20 Jul 2026cs.AIcited by 0

On the map: Retrieval-augmented generation → RAG robustness and adaptive retrieval

Abstract (arXiv)

In retrieval-augmented generation (RAG), semantic relevance asks whether a source matches a query in meaning, while answerability asks whether it contains sufficient information to answer the query. A Semantic-Answerability Gap (SAG) may arise in retrieval when a retriever can reach semantically relevant sources yet fail to identify those that are uniquely answerable. We uncover this gap using tables as a controlled setting, where shared schemas and entities provide strong semantic signals while localized content and row-column bindings distinguish answerable from non-answerable sources. Using TCR-Bench, a controlled sibling-table benchmark, we find that dense retrievers achieve only 18.2% top-1 target retrieval, reducing QA F1 from 0.755 with the oracle table to 0.330 with retrieved top-5 tables. Controlled diagnostics show that retrievers favor semantic volume over sufficiency, respond weakly to row-column binding disruptions, and struggle to distinguish Targets from Siblings. Explicit answerability assessment substantially improves target identification, while fine-tuning shows that answerability is learnable but difficult to transfer without compromising broad semantic retrieval.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in RAG robustness and adaptive retrieval, one of 15 directions in Retrieval-augmented generation. Builds on 51 of its 64 references that are in the map. First posted on arXiv 20 Jul 2026.

Builds on

51 of 64 references.

Closest ancestors · 9

Same direction, newest first

  1. 2026Beyond Semantic Relevance: Counterfactual Risk Minimization for Robust Retrieval-Augmented Generation2 May
  2. 2025ARK: Answer-Centric Retriever Tuning via KG-augmented Curriculum Learning20 Nov
  3. 2025Fact or Facsimile? Evaluating the Factual Robustness of Modern Retrievers28 Aug
  4. 2025Pneuma: Leveraging LLMs for Tabular Data Representation and Retrieval in an End-to-End System12 Apr
  5. 2025RAG over Tables: Hierarchical Memory Index, Multi-Stage Retrieval, and Benchmarking2 Apr
  6. 2025Collapse of Dense Retrievers: Short, Early, and Literal Biases Outranking Factual Evidence6 Mar
  7. 2024RE-RAG: Improving Open-Domain QA Performance and Interpretability with Relevance Estimator in Retrieval-Augmented Generation9 Jun
  8. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection17 Oct
  9. 2022End-to-End Table Question Answering via Retrieval-Augmented Generation30 Mar

Other references · 42

  1. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,160
  2. 2020Dense Passage Retrieval for Open-Domain Question Answeringcited by 3,779
  3. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
  4. 2021BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Modelscited by 1,225
  5. 2025Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Modelscited by 1,175
  6. 2020ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERTcited by 1,160
  7. 2018Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Taskcited by 1,120
  8. 2024M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillationcited by 1,050
  9. 2022Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questionscited by 911
  10. 2023Towards General Text Embeddings with Multi-stage Contrastive Learningcited by 705
  11. 2023Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLscited by 667
  12. 2015Compositional Semantic Parsing on Semi-Structured Tablescited by 509
30 more
  1. 2022Precise Zero-Shot Dense Retrieval without Relevance Labelscited by 508
  2. 2023Making Retrieval-Augmented Language Models Robust to Irrelevant Contextcited by 364
  3. 2021SPLADE: Sparse Lexical and Expansion Model for First Stage Rankingcited by 283
  4. 2020HybridQA: A Dataset of Multi-Hop Question Answering over Tabular and Textual Datacited by 263
  5. 2021SPLADE v2: Sparse Lexical and Expansion Model for Information Retrievalcited by 126
  6. 2023Table Meets LLM: Can Large Language Models Understand Structured Table Data? A Benchmark and Empirical Studycited by 116
  7. 2024Jasper and Stella: distillation of SOTA embedding modelscited by 109
  8. 2025On the Theoretical Limitations of Embedding-Based Retrievalcited by 98
  9. 2024A Survey on Retrieval-Augmented Text Generation for Large Language Modelscited by 77
  10. 2021Open Domain Question Answering over Tables via Dense Retrievalcited by 74
  11. 2023Tabular Representation, Noisy Operators, and Impacts on Table Structure Understanding Tasks in LLMscited by 49
  12. 2020The Curse of Dense Low-Dimensional Information Retrieval for Large Index Sizescited by 48
  13. 2023Rethinking Tabular Data Understanding with Large Language Modelscited by 34
  14. 2024Is Table Retrieval a Solved Problem? Exploring Join-Aware Multi-Table Retrievalcited by 27
  15. 2025TableRAG: A Retrieval Augmented Generation Framework for Heterogeneous Document Reasoningcited by 22
  16. 2018Ad Hoc Table Retrieval using Semantic Similaritycited by 21
  17. 2025TARGET: Benchmarking Table Retrieval for Generative Taskscited by 17
  18. 2025HoH: A Dynamic Benchmark for Evaluating the Impact of Outdated Information on Retrieval-Augmented Generationcited by 13
  19. 2022Table Retrieval May Not Necessitate Table-specific Model Designcited by 11
  20. 2025Positional Biases Shift as Inputs Approach Context Window Limitscited by 9
  21. 2024Tabular Embedding Model (TEM): Finetuning Embedding Models For Tabular RAG Applicationscited by 8
  22. 2025Tailoring Table Retrieval from a Field-aware Hybrid Matching Perspectivecited by 7
  23. 2024W-RAG: Weakly Supervised Dense Retrieval in RAG for Open-domain Question Answeringcited by 6
  24. 2025mmRAG: A Modular Benchmark for Retrieval-Augmented Generation over Text, Tables, and Knowledge Graphscited by 5
  25. 2025Dynamic Context Selection for Retrieval-Augmented Generation: Mitigating Distractors and Positional Biascited by 3
  26. 2025Mixture-of-RAG: Integrating Text and Tables with Large Language Modelscited by 2
  27. 2025CRAFT: Training-Free Cascaded Retrieval for Tabular QAcited by 2
  28. 2025TableCopilot: A Table Assistant Empowered by Natural Language Conditional Table Discoverycited by 2
  29. 2026Scaling Laws for Embedding Dimension in Information Retrievalcited by 2
  30. 2026UsefulBench: Towards Decision-Useful Information as a Target for Information Retrievalcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Unanswerability Evaluation for Retrieval Augmented Generation
  2. 2026A Hierarchical Consistency Framework for Auditing Retrieval-Augmented Generation Systems
  3. 2025RAGSmith: A Framework for Finding the Optimal Composition of Retrieval-Augmented Generation Methods Across Datasets
  4. 2026EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering
  5. 2025Investigating the Robustness of Retrieval-Augmented Generation at the Query Level
  6. 2026DF-RAG: Query-Aware Diversity for Retrieval-Augmented Generation
  7. 2024Semantic Tokens in Retrieval Augmented Generation
  8. 2026CQC-RAG: Robust Retrieval-Augmented Generation via Cross-Query Consistency