Datasets:

Languages:
Japanese
Tags:
Not-For-All-Audiences
License:

Accept the Dataset Access Agreement

By clicking on 'Accept Agreement' below, you are accessing this dataset and accepting the UncommonCrawl Dataset Access Agreement, which contains an arbitration agreement.

UncommonCrawl Dataset Access Agreement

Last Updated: 2025-05-25

Portions copyright © 2025 The UncommonCrawl Project. All rights reserved.

This dataset access agreement is an agreement between you and the UncommonCrawl Project (an unincorporated nonprofit association based in Wyoming, USA).

THIS AGREEMENT CONTAINS A PROVISION FOR MANDATORY BINDING ARBITRATION. FOR MORE INFORMATION, SEE SECTION 6 ("DISPUTE RESOLUTION") AND READ IT CAREFULLY.

1. Definitions

"Agreement" means this dataset access agreement.

"Dataset" means the data and other resources provided alongside this agreement.

"Curator" means the people who created, developed, distributed, or contributed to this Dataset.

2. Acceptance

By accessing, receiving, or using this Dataset, you accept this Agreement.

3. Grant of Rights

Access to this Dataset is provided for all lawful purposes, as long as you comply with this Agreement.

4. Conditions

(a) You must include this Agreement and the above copyright notice in all copies of this Dataset, and you must ensure that anyone who accesses, receives, or uses this Dataset accepts this Agreement.

(b) You give up all claims (copyright, patent, contract, or any legal claim) against the Curators for anything related to this Dataset or its development.

(c) You grant the Curators any rights you hold that are necessary to develop and distribute this Dataset.

(d) You must not attempt to use this Dataset to generate negative publicity for the Curators.

(e) You must not knowingly distribute copies of this Dataset to known bad actors who will breach this Agreement.

5. Disclaimer

This Dataset is provided "AS IS" without any warranty, express or implied. The Curators will not be liable for any damages related to this Dataset under any kind of legal claim. You will be liable to the Curators for any damages which result from you failing to comply with this Agreement.

6. Dispute Resolution

(a) This Agreement is governed by the laws of the State of Wyoming without regard to conflict of laws principles or the United Nations Convention on Contracts for the International Sale of Goods.

(b) Any dispute related to this Agreement or this Dataset must be resolved by binding arbitration through AirJudge (https://airjudge.me/rules), whose rules are incorporated by reference into this Agreement. If AirJudge refuses the dispute, then the state and federal courts in Arizona have exclusive jurisdiction.

7. Other Provisions

(a) This Agreement is the entire agreement and replaces all prior agreements.

(b) If you breach this Agreement and do not fix it within 14 days of notice, then this Agreement ends. Sections 1, 4, 5, 6, and 7 survive the end of this Agreement.

(c) If any part of this Agreement is unenforceable, then the rest remains in effect, except that if Section 4(b) or 4(c) is unenforceable, then the entire Agreement ends.

(d) The English version of this Agreement controls over any translations.

Log in or Sign Up to review the conditions and access this dataset content.

Dataset Card for Syosetu-2505-Raw

This dataset is a collection of WARC files containing ~1,319,364 novels and accompanying metadata scraped from Syosetuka ni Narou (https://syosetu.com) in May 2025.

Dataset Details

Uses

This dataset is intended to be processed and used in the training of large language models.

Dataset Structure

The dataset consists WARC files containing only "resource" records. The URLs fetched are a novel's main page, the info page ("infotop"), and the chapters, if any.

Dataset Creation

Curation Rationale

We provide this data to improve the generation of creative works in Japanese by large language models.

Source Data

The source data is Japanese web novels from https://syosetu.com.

Who are the source data producers?

The authors on Syosetuka ni Narou.

Personal and Sensitive Information

Generally public webnovels should not contain personal information, so we will assume there is no personal information contained within.

Bias, Risks, and Limitations

This dataset consists solely of works of fiction produced for entertainment purposes. The text is not intended to be an accurate reflection of reality. Additionally, this dataset contains novels which are "not safe for work" (18+).

Contact

The UncommonCrawl Project can be contacted by email at [email protected].

Downloads last month
79

Collection including UncommonCrawl/syosetu-2505-raw