UDA FETA-QA (orgrctera/uda_feta_qa)

Overview

This dataset is the FETA-QA (FeTaQA-style) slice of the UDA (Unstructured Document Analysis) benchmark: 1,023 question–answer instances over Wikipedia-sourced tables, packaged for retrieval-oriented evaluation in RAG pipelines.

UDA (Hui et al., NeurIPS 2024) is a benchmark suite for Retrieval-Augmented Generation (RAG) over messy, real-world documents (PDF/HTML) where evidence mixes narrative text and structured content. In the Wikipedia track, UDA includes FetaTab—examples aligned with FeTaQA (free-form table QA) at the scale of this release (878 documents and 1,023 Q&A pairs in the UDA paper’s configuration for this subset).

FeTaQA (Nan et al., TACL 2022) introduces free-form answers to questions over Wikipedia tables: systems must retrieve and reason over tabular evidence and produce natural-language explanations, not only short spans. UDA adopts FeTaQA-style supervision within its broader document-analysis benchmark so that parsing, chunking, and retrieval are evaluated alongside generation.

In this Hub release, each row is a retrieval task instance: given a question (input), a system must locate the right document/table context and ground a free-form answer—consistent with the UDA setting, where retrieval is a first-class step before answering.

Task

• Task type: Retrieval (within a RAG / document-analysis pipeline) for FETA-QA / FeTaQA-style table question answering over Wikipedia-derived tables.
• Input: A natural-language question (input) about entities, events, or facts described in a table (and optionally surrounding article context in a full ingestion pipeline).
• Supervision / reference: expected_output is a JSON string with a gold free-form answer (answers), document URL (doc_url), and structured evidence (evidence: table_array, highlighted_cell_ids supporting the answer). metadata records UDA identifiers (sub_benchmark: feta_qa).

Evaluation typically combines retrieval quality (whether the correct table or cells are retrieved) with answer quality (fluency and factual alignment with gold free-form text and highlighted cells), following FeTaQA and UDA protocols.

Background

FeTaQA (FETA-QA)

FeTaQA targets question answering over tables where answers are free-form text that may integrate multiple table cells. Compared with short-answer table QA, it stresses retrieval, inference, and surface realization from structured sources. Annotations include supporting cells, enabling training and evaluation of both retrieval and generation.

UDA benchmark

UDA revisits RAG and LLM-based document analysis across domains using 2,965 real-world documents and 29,590 expert-annotated Q&A pairs, with sources kept in original formats to stress parsing and retrieval as well as generation. The FetaTab Wikipedia subset corresponds to this dataset’s 1,023 examples in the default split.

Data fields

Column	Type	Description
input	string	Question text posed over the table (and article, in full-document settings).
expected_output	string	JSON string with answers (gold free-form text), doc_url, and evidence (table_array, highlighted_cell_ids).
metadata	struct	benchmark_name (uda_feta_qa), benchmark_type (uda), split, sub_benchmark (feta_qa), and value (JSON string with identifiers such as label_key, label_file, doc_url, q_uid).

Splits: Single split default with 1,023 examples.

Examples

The following rows are taken from the dataset (JSON in expected_output is shown formatted for readability; large table_array values are abbreviated with [...]).

Example 1 — theatre and production

• input: When and in what play did Platt appear at the Music Box Theatre?
• expected_output:

{
  "answers": "In 2016 and 2017, Platt played in Dear Evan Hansen on Broadway at the Music Box Theatre.",
  "doc_url": "http://en.wikipedia.org/wiki/Ben_Platt_(actor)",
  "evidence": {
    "table_array": [
      ["Year", "Production", "Role", "Venue", "Notes"],
      ["2002", "The Music Man", "Winthrop Paroo", "Hollywood Bowl", "Los Angeles"],
      "...",
      ["2016–2017", "Dear Evan Hansen", "Evan Hansen", "Music Box Theatre", "Broadway"]
    ],
    "highlighted_cell_ids": [[10, 1], [11, 0], [11, 1], [11, 3], [11, 4]]
  }
}

Example 2 — role at a venue

• input: Which role did Ben Platt play at the Second Stage Theatre in 2016?
• expected_output:

{
  "answers": "In 2016, Platt played the role of Evan Hansen at Second Stage Theatre.",
  "doc_url": "http://en.wikipedia.org/wiki/Ben_Platt_(actor)",
  "evidence": {
    "table_array": [
      ["Year", "Production", "Role", "Venue", "Notes"],
      "...",
      ["2016", "Dear Evan Hansen", "Evan Hansen", "Second Stage Theatre", "Off-Broadway"]
    ],
    "highlighted_cell_ids": [[9, 0], [10, 0], [10, 2], [10, 3], [11, 2]]
  }
}

metadata.value (structure, example):

{
  "label_key": "Ben Platt (actor)",
  "label_file": "feta_qa",
  "doc_url": "http://en.wikipedia.org/wiki/Ben_Platt_(actor)",
  "q_uid": "14717"
}

References

FeTaQA (source task & data lineage)

Linyong Nan, Chiachun Hsieh, Ziming Mao, Xi Victoria Lin, Neha Verma, Rui Zhang, Wojciech Kryściński, Nick Schoelkopf, Riley Kong, Xiangru Tang, Murori Mutuma, Ben Rosand, Isabel Trindade, Renusree Bandaru, Jacob Cunningham, Caiming Xiong, Dragomir Radev. FeTaQA: Free-form Table Question Answering. Transactions of the Association for Computational Linguistics (TACL), 2022.

• Abstract (short): Introduces FeTaQA with 10K Wikipedia-based {table, question, free-form answer, supporting table cells} pairs; free-form answers require retrieval, inference, and integration from structured data—more challenging than short-fact table QA. Provides pipeline and end-to-end baselines.
• arXiv: https://arxiv.org/abs/2104.00369
• ACL Anthology: https://aclanthology.org/2022.tacl-1.3/
• Project / code: https://github.com/Yale-LILY/FeTaQA

UDA benchmark (suite containing this FETA-QA slice)

Yulong Hui, Yao Lu, Huanchen Zhang. UDA: A Benchmark Suite for Retrieval Augmented Generation in Real-world Document Analysis. NeurIPS 2024 (Datasets and Benchmarks Track).

• Abstract (short): Presents UDA with thousands of real-world documents and tens of thousands of expert-annotated Q&A pairs; evaluates LLM- and RAG-based document analysis and highlights parsing and retrieval design choices alongside generation.
• arXiv: https://arxiv.org/abs/2406.15187
• NeurIPS proceedings: https://proceedings.neurips.cc/paper_files/paper/2024/hash/7c06759d1a8567f087b02e8589454917-Abstract-Datasets_and_Benchmarks_Track.html
• Code & resources: https://github.com/qinchuanhui/UDA-Benchmark

Related Hub resources

• Aggregated UDA QA (reference): qinchuanhui/UDA-QA

Citation

If you use this dataset, please cite both FeTaQA and UDA (and this dataset record as appropriate):

@article{nan-etal-2022-fetaqa,
  title   = {FeTaQA: Free-form Table Question Answering},
  author  = {Nan, Linyong and Hsieh, Chiachun and Mao, Ziming and Lin, Xi Victoria and Verma, Neha and Zhang, Rui and Kry{\'s}ci{\'n}ski, Wojciech and Schoelkopf, Nick and Kong, Riley and Tang, Xiangru and Mutuma, Murori and Rosand, Ben and Trindade, Isabel and Bandaru, Renusree and Cunningham, Jacob and Xiong, Caiming and Radev, Dragomir},
  journal = {Transactions of the Association for Computational Linguistics},
  year    = {2022},
  volume  = {10},
  pages   = {35--49}
}

@article{hui2024uda,
  title   = {UDA: A Benchmark Suite for Retrieval Augmented Generation in Real-world Document Analysis},
  author  = {Hui, Yulong and Lu, Yao and Zhang, Huanchen},
  journal = {arXiv preprint arXiv:2406.15187},
  year    = {2024}
}

License

Use this dataset in compliance with the original FeTaQA and UDA data licenses and terms. The FeTaQA project (CC BY-SA 4.0 for the original release) and UDA documentation specify conditions; verify terms for your use case before redistribution or commercial use.