Home/Language & Culture/Korean Document Ambiguity and Implicit Meaning Understanding Benchmark Dataset
Korean Document Ambiguity and Implicit Meaning Understanding Benchmark Dataset
#ambiguity detection #implicit meaning reasoning #semantic interpretation #Korean document understanding #ambiguity interpretation #implicit meaning comprehension
- 5000
- 0.2GB
- JSON
- CATL
MOBIUSI INC
Updated:2026-10-19
AI Analysis & Value Prop
This benchmark covers challenging phenomena in Korean documents, including omitted subjects, honorific tone, negation, conditional exceptions, ambiguous references, polysemy, and polite refusals. Its JSON records organize the source document, ambiguity points, candidate interpretations, correct semantics, and semantic parses. The examples support domain-specific supervised fine-tuning and evaluation of document understanding, helping identify gaps in contextual and pragmatic reasoning.
Dataset Insights
Sample Examples
{}
Technical Specifications
| Field | Type | Description |
|---|---|---|
| file_name | string | File name |
| raw_document | object | The original Korean document before semantic analysis, along with basic contextual information. |
| semantic_parse | object | A structured analysis of challenging language phenomena that explains the reasoning from the source text to the correct meaning. |
| ambiguity_points | array | Locations in the source text involving omissions, negation, conditions, references, polysemy, or pragmatics, with explanations of the ambiguity. |
| correct_semantics | object | The intended meaning and communicative function determined from the context and Korean pragmatic conventions. |
| candidate_interpretations | array | Possible semantic interpretations of the document or its key ambiguity points, with assessments of supporting evidence. |
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
- Which Korean language phenomena are covered?
- The examples focus on challenging aspects of document meaning, including omitted subjects, honorific tone, negation, conditional exceptions, ambiguous references, polysemy, and polite refusals, with explanations and semantic analyses.
- What format and fields does the data include?
- The data is organized as JSON, with fields for the source document, ambiguity points, candidate interpretations, correct semantics, and semantic parse. This structure supports field-level inspection, filtering, and integration into data workflows.
- How can I obtain the data, and what licensing applies?
- Refer to the product listing for the available acquisition method, delivery details, and license terms. Confirm that the license covers your intended use, especially for commercial training or redistribution.
- Which training and evaluation workflows is it suited for?
- It supports domain-specific supervised fine-tuning of Korean language models and evaluation of document understanding, semantic disambiguation, coreference resolution, and pragmatic inference. It can help assess how well models use context to interpret implied meaning.
Can't find the data you need?
Post a request and let data providers reach out to you.
Get this Dataset
Verified for Enterprise UseCite this Work
@dataset{Mobiusi2026,
title={Korean Document Ambiguity and Implicit Meaning Understanding Benchmark Dataset},
author={MOBIUSI INC},
year={2026},
url={https://www.mobiusi.com/datasets/c9809c8579aa04dc0039e38ecfcfa3a2?dataset_scene_id=21},
urldate={2026-10-16},
keywords={Korean document understanding,Korean ambiguity dataset,Korean semantic understanding,Korean implicit meaning,Korean coreference resolution,Korean pragmatic inference,Korean SFT data,language model evaluation},
version={1.0}
}Using this in research? Please cite us.
placeholder
placeholder
placeholder
placeholder
placeholder
placeholder
placeholder