EduAssess Controlled Non-Commercial Research Use Terms 1.0 (license: other). The complete controlling text appears in LICENSE and is reproduced verbatim at the end of this card.

Terms

Access requires prior manual approval by the dataset owner. Approved users may use the dataset only for the specific non-commercial research project stated in their access request. Permission is personal, project-specific, non-transferable, revocable, and limited to the approved research period.

Redistribution, sublicensing, mirroring, credential sharing, and publication of the records or a substantial reconstruction are prohibited. Commercial, production, consulting, paid-service, advertising, and product-development use requires separate prior written authorization from the data owner. These paragraphs summarize the license and do not replace the controlling terms reproduced below.

Data provenance and transformation

The released records were constructed separately for controlled research use. Institution, program, discipline, and indicator identifiers are synthetic and do not provide a mapping to real entities. Values, derived statistics, chart payloads, and English reference texts in this repository belong to the released records and must not be represented as measurements of real institutions.

The original institutional records, institution and program names, expert reports, prompts, model outputs, and private experimental datasets are not distributed. Each released sample is subject to a minimum-cell release rule: samples with fewer than five synthetic entity-level observations are excluded. The release does not claim k-anonymity, l-diversity, entity perturbation, local differential privacy, or an epsilon-differential-privacy guarantee.

The dataset is not intended for operational evaluation or decision-making about real institutions, programs, or people.

data/controlled_dataset.jsonl: UTF-8 JSON Lines records.
schema.json: machine-readable schema summary.
schema.md: field-level definitions and consistency constraints.
splits/public_synthetic_evaluation_ids.txt: authoritative ordered ID list for the public_synthetic_evaluation split.
LICENSE: controlling license terms.
ACCESS_POLICY.md: application and manual-approval procedure.
release_manifest.json: release status, date, contact, and policy metadata.

Each non-empty split-list line identifies exactly one JSONL record. Every JSONL record must occur exactly once in the list, in the same order.

Schema

Field	Type	Description
`sample_id`	string	Unique synthetic sample identifier.
`split`	string	Dataset split identifier.
`synthetic`	boolean	Indicates that the example is synthetic.
`task`	object	Discipline identifier, synthetic year, chapter, and language.
`indicator`	object	Synthetic indicator ID, English name, and unit.
`records`	list of objects	Synthetic institution/program observations and values.
`statistics`	object	Count, mean, extrema, and corresponding entity IDs derived from `records`.
`chart`	object	Chart type, title, labels, values, and unit aligned with `records`.
`reference_report_en`	string	English reference text grounded in the synthetic records.

See schema.md for nested fields, required types, and validation rules.

Validation

Before upload, the release package is checked for valid JSONL, unique sample IDs, English-only content, synthetic identifier formats, minimum cell size, and exact consistency among records, statistics, chart labels, and chart values. Validation of the public release is separate from the private experimental pipeline and does not establish distributional equivalence to the private experimental datasets.

Citation

Until the associated paper receives its final bibliographic record, cite the dataset repository as follows:

@misc{eduassess_research_dataset_2026,
  title        = {EduAssess Research Dataset},
  author       = {{Anonymous Authors}},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/Alli77/eduassess-research-dataset}},
  note         = {Controlled access; non-commercial research use only}
}

The citation will be updated when the associated paper and, if applicable, a persistent dataset identifier become available.

Dataset citation is required for publications using the released material. Users are also encouraged to notify the repository administrators of substantive derivative research, although notification does not replace any license obligation.

Maintenance, corrections, and withdrawal

During double-blind review, maintenance questions, correction requests, security reports, access revocation, and dataset-withdrawal requests should be submitted through the repository's Hugging Face Discussions page. Author names, institutional affiliations, and direct contact details are intentionally withheld during review.

Please include the Hugging Face username and the approved project name when the request concerns an existing access grant. Do not include private source data in discussion messages.

License terms

The text below is identical to the controlling LICENSE file. If a rendering or transcription discrepancy is discovered, distribution must be paused until the two copies are made identical.

EduAssess Controlled Non-Commercial Research Use Terms 1.0

No automatic license Possession of a copy, submission of an access request, or visibility of dataset metadata does not grant permission to use the dataset. Permission begins only after the Licensor manually approves the named applicant's request.
Limited research permission Subject to continued compliance with these terms, an approved applicant may use the dataset solely for the non-commercial research project described in the approved request. The permission is personal, non-exclusive, non-transferable, revocable, and limited to the approved project and research period.
Non-commercial restriction The dataset must not be used for commercial, production, consulting, paid service, advertising, product-development, or other revenue-generating purposes. Research funded by or performed for a commercial entity requires separate prior written authorization from the Licensor.
No redistribution or access sharing The applicant must not publish, redistribute, sublicense, mirror, sell, lend, or otherwise provide the dataset or any substantial reconstruction of it to another person or organization. Access credentials must not be shared. Each collaborator must request and receive separate approval.
Privacy and representation safeguards The applicant must not attempt to link synthetic identifiers or values to real institutions, programs, people, or private records. The dataset must not be represented as measurements of real entities or used to make decisions about real entities.
Derived materials Research results, aggregate statistics, and trained models may be published only when they do not reproduce dataset records or enable reconstruction of a substantial portion of the dataset. Redistribution of modified, translated, reformatted, or augmented dataset records is prohibited without separate prior written authorization.
Security and deletion The applicant must use reasonable safeguards to prevent unauthorized access and must promptly report suspected disclosure to the Licensor. On expiration, revocation, or request by the Licensor, the applicant must stop use and delete all dataset copies, except where retention is required by law.
Attribution Any publication using the dataset must identify it by name, cite the associated paper or repository citation when available, and state that access was granted for non-commercial research. Attribution does not imply endorsement.
No warranty; reservation of rights The dataset is provided "AS IS", without warranties or guarantees of accuracy, fitness, availability, or non-infringement. The Licensor reserves all rights not expressly granted and may approve, deny, suspend, or revoke access.
Additional authorization Any use not expressly permitted above requires separate prior written authorization from the Licensor. Approval for one project or applicant does not authorize another project, applicant, institution, or use.

Downloads last month: 17

Total file size:

170 kB

Request access for non-commercial research

EduAssess Research Dataset

License