Passage-Based Question Answering Evaluation Dataset

#passage-based question answering #evidence identification #semantic understanding #reading comprehension #educational assessment #evidence-based question answering

5000
0.3GB
JSON
CATL
MOBIUSI INC

Updated:2026-10-19

AI Analysis & Value Prop

This dataset contains web-text questions from digital question banks that require answers grounded in short passages, case studies, chart descriptions, or reference materials. It includes raw question records, normalized reference answers, annotated answer evidence, and semantic annotations, supporting evaluation of contextual evidence use, reference resolution, and answer grounding. It is suitable for question answering benchmarks, reading comprehension analysis, and the evaluation and improvement of education-focused models.

Dataset Insights

Sample Examples

{}

Technical Specifications

Field	Type	Description
file_name	string	File name
reference_answer	string	A cleaned and normalized version of the original answer, provided as a reference for evaluating model responses.
raw_question_record	object	The unnormalized source record for a passage-based question, including the supplied material, question, options, and original answer where available.
evidence_annotations	array	Identifies material spans that support the reference answer, their character positions, and their evidential roles.
semantic_annotations	object	Annotations derived from the material, question, and reference answer, covering question type, reasoning requirements, reference relations, and answer support.

Compliance Statement

Authorization Type	Proprietary - Commercial AI Training License (No Redistribution)
Commercial Use	Requires exclusive subscription or authorization contract (monthly or per-invocation charging)
Privacy and Anonymization	No PII, no real company names, simulated scenarios follow industry standards
Compliance System	Compliant with China's Data Security Law / EU GDPR / supports enterprise data access logs

Frequently Asked Questions

What types of questions are included?: The dataset contains web-text questions that must be answered using provided materials, including short passages, case studies, chart descriptions, and reference excerpts. It focuses on understanding and applying information from context.

What annotations are provided for each item?: The fields include a raw question record, a normalized reference answer, answer evidence annotations, and semantic annotations. Evidence annotations identify supporting text spans and their character positions, while semantic annotations can describe question type, reasoning requirements, reference relations, and answer support.

Which model capabilities can this dataset evaluate?: It can evaluate reading comprehension, passage-based question answering, contextual evidence localization, reference resolution, and answer grounding. It can also support analysis of education-focused question answering systems.

How can the dataset be obtained, and what are the licensing terms?: For availability, formats, and licensing terms, refer to the dataset product page or contact the provider. Before use, verify the permitted purposes and any restrictions on the original question content.

Can't find the data you need?

Post a request and let data providers reach out to you.

Get this Dataset

Verified for Enterprise Use

Cite this Work

@dataset{Mobiusi2026,
  title={Passage-Based Question Answering Evaluation Dataset},
  author={MOBIUSI INC},
  year={2026},
  url={https://www.mobiusi.com/datasets/bef028683c68c309c2678257b6aa1e93?dataset_scene_id=7},
  urldate={2026-10-15},
  keywords={passage-based questions,reading comprehension dataset,question answering evaluation,educational QA data,context understanding,answer evidence annotation,coreference resolution,digital question bank,semantic understanding benchmark},
  version={1.0}
}

Using this in research? Please cite us.

placeholder

Products

Scene

Domain

Modality

Task