Employment Dispute Evidence Retrieval and Source-Grounded QA Benchmark

#legal evidence retrieval #source-grounded question answering #answerability assessment #employment dispute resolution #labor law compliance #legal evidence analysis

6300
0.4GB
JSON
CATL
MOBIUSI INC

Updated:2026-11-05

AI Analysis & Value Prop

This benchmark contains question-answering examples about employment dispute case facts, agreements, rules, and decisions. Models must retrieve relevant support from the supplied materials and answer with clear distinctions between established facts and unconfirmed details. Each example captures the question, source materials, grounded answer, answerability status, evidence references, and unconfirmed points for evaluating and improving retrieval-augmented legal systems.

Dataset Insights

Sample Examples

{}

Technical Specifications

Field	Type	Description
file_name	string	File name
answer	string	An answer based on the supplied materials that distinguishes established facts from uncertainty where relevant.
evidence	array	Specific source passages and their relationship to the answer.
question	string	The employment dispute question to be answered using the provided source materials.
source_materials	array	The case facts, agreements, rules, or decisions supplied as evidence for answering the question.
unconfirmed_points	array	Claims or details relevant to the question that the supplied materials do not establish or allow the answer to verify.
answerability_status	string	The extent to which the supplied materials establish an answer to the question.

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 benchmark include?: Examples pair employment dispute questions with source materials such as case facts, agreements, rules, or decisions. Fields include a grounded answer, answerability status, evidence references, and points the materials do not confirm.

Which capabilities can it evaluate?: It can assess whether a model retrieves relevant support from provided materials, cites sources accurately, distinguishes established facts from unconfirmed details, and avoids unsupported conclusions when evidence is insufficient.

What are the dataset size and quality specifications?: The current description does not specify the sample count or quantitative quality metrics. The examples are organized around questions, sources, answers, and evidence to support RAG evaluation and error analysis. Check the product page for current size and version details.

How can I access the data, and what are the licensing terms?: Consult the dataset product page for available formats, access instructions, and applicable license terms. Usage rights are governed by the license published for the product.

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={Employment Dispute Evidence Retrieval and Source-Grounded QA Benchmark},
  author={MOBIUSI INC},
  year={2026},
  url={https://www.mobiusi.com/datasets/1eec1a6407335869041b19a78d155380},
  urldate={2026-10-21},
  keywords={employment dispute dataset,employment law question answering,legal evidence retrieval,RAG evaluation benchmark,open-book QA benchmark,legal QA dataset,evidence citation evaluation,labor dispute AI},
  version={1.0}
}

Using this in research? Please cite us.

placeholder

Scene

Domain

Modality

Task