Negation and Exception Question-Answering Challenge Set
- Records
- 5000
- Size
- 0.4GB
- Format
- JSON
- Provider
MOBIUSI INC- Updated
- v1.02026-11-04
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.
Overview
Dataset Insights
Preview
Sample Examples
Schema
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. |
Trust
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
FAQ
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
High-quality data, ready for your next AI breakthrough.
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?dataset_scene_id=7},
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.