Negation and Exception Question-Answering Challenge Set

#question answering #negation understanding #exception and scope reasoning #exam question understanding #educational assessment #semantic scope analysis

5000
0.4GB
JSON
CATL
MOBIUSI INC

Updated:2026-11-04

AI Analysis & Value Prop

This dataset contains text-based question-answer samples collected from online question banks, covering negation, exceptions, scope, and conditional constraints such as “incorrect,” “does not belong to,” “except,” and “main reason.” It includes normalized questions, annotations of limiting expressions, option-level semantic analysis, gold answers, and diagnostic labels for semantic failure modes. Use it to evaluate question-bank QA, semantic understanding, and model robustness in exam settings.

Dataset Insights

Sample Examples

{}

Technical Specifications

Field	Type	Description
file_name	string	File name
raw_record	object	Original text collected from a question-bank webpage, including structured question, option, and answer content.
gold_answer	object	A normalized annotation of the correct answer and its semantic basis.
cue_annotations	array	Negation, exception, scope, and conditional cues in the question that affect answer selection.
option_analysis	array	An assessment of how each option matches the question constraints and the requirements for the correct answer.
question_task_type	string	The primary negation, exception, scope, or conditional constraint assessed by the question.
diagnostic_annotation	object	Labels for evaluating whether a model correctly handles negation, scope, and conditional constraints in a question.
normalized_question_text	string	Question text with formatting noise cleaned up without changing its meaning.

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 does the dataset contain?: It is based on text question-answer samples from online question banks and includes raw question content, normalized questions, constraint-type and cue annotations, option-level semantic analysis, gold answers, and diagnostic information.

Which capabilities can it evaluate?: The dataset can assess how well models interpret negation, exceptions, scope limitations, and conditional constraints. It also supports question-bank QA, answer-option matching, and analysis of related semantic failure modes.

What annotation and text quality information is available?: The data provides raw samples and normalized question text, with structured fields for constraint cues, answer rationale, and diagnostic labels. Formatting and completeness may vary across source webpages, so sampling and inspection are recommended for the intended use.

How can I access the data, and what are the licensing terms?: Check the dataset product page for available versions, access methods, and license terms. Use of content originating from webpages must follow the stated license and applicable regulations; verify that the terms cover your intended use.

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={Negation and Exception Question-Answering Challenge Set},
  author={MOBIUSI INC},
  year={2026},
  url={https://www.mobiusi.com/datasets/8faec372ffefda8b583f707f7c67aa18},
  urldate={2026-10-15},
  keywords={negation QA dataset,exception question data,question bank evaluation,negation understanding,scope constraint detection,conditional reasoning data,educational AI,semantic robustness evaluation},
  version={1.0}
}

Using this in research? Please cite us.

placeholder

Scene

Domain

Modality

Task