Patent Novelty and Single-Document Prior-Art Evidence QA Dataset
#patent novelty verification #claim-feature mapping #evidence-based question answering #patent novelty assessment #prior art analysis #patent examination support
- 5000
- 0.4GB
- JSON
- CATL
MOBIUSI INC
Updated:2026-10-19
AI Analysis & Value Prop
This dataset presents novelty questions based on patent claims and a single prior-art document, with fields for claim text, source evidence, questions, conclusions, citations, and feature correspondence analysis. It focuses on whether one document discloses the relevant claimed features and requires answers to connect evidence to those features using only the supplied materials, while noting material differences or missing limitations. It supports domain-specific model fine-tuning and evaluation for patent invalidity analysis, evidence retrieval, and novelty assessment.
Dataset Insights
Sample Examples
{}
Technical Specifications
| Field | Type | Description |
|---|---|---|
| file_name | string | File name |
| claim_text | string | The full text of the patent claim being assessed. |
| answer_text | string | A concise answer stating the conclusion and grounding it in the cited evidence and feature correspondence. |
| question_text | string | The novelty question asking whether the prior-art document discloses the relevant features of the claim. |
| cited_evidence | array | Specific passages from the prior-art document that support the assessment, with a locator and the claim feature addressed by each passage. |
| prior_art_text | string | The text of the single prior-art document used as the evidence source for the assessment. |
| anticipation_conclusion | string | The assessment of whether the single prior-art document discloses the claimed subject matter for novelty purposes. |
| correspondence_analysis | string | An explanation of how each cited passage corresponds to a claim feature, including material differences or missing limitations. |
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 organized around patent claims and a single prior-art document, with novelty questions, assessment conclusions, specific cited passages, feature correspondence analysis, and final answers. The fields link each question to its evidence and conclusion.
- How does it support novelty-assessment training?
- Examples require answers grounded in the supplied materials, mapping passages from the prior-art document to relevant claim features and noting material differences or undisclosed limitations. This supports training and evaluation of evidence-grounded question answering and novelty analysis.
- Can the cited evidence be verified?
- The dataset is designed to include specific passages and locator information from the prior-art document, along with the claim feature addressed by each passage. Verification depends on the completeness of the supplied document text and locator details.
- How can I obtain the dataset and confirm its license?
- Check the product page or the data provider's latest information for availability, formats, and licensing terms. Confirm permissions for the dataset and relevant patent documents before commercial training, redistribution, or use in legal practice.
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={Patent Novelty and Single-Document Prior-Art Evidence QA Dataset},
author={MOBIUSI INC},
year={2026},
url={https://www.mobiusi.com/datasets/9682b357209d38b3f15d7dcbb5435988?dataset_scene_id=8},
urldate={2026-10-17},
keywords={patent novelty dataset,patent invalidity QA,prior art evidence,patent claim analysis,evidence citation dataset,patent domain fine-tuning,novelty assessment model,grounded legal QA},
version={1.0}
}Using this in research? Please cite us.
placeholder
placeholder
placeholder
placeholder
placeholder
placeholder
placeholder