This dataset provides a collection of practical, multilingual and localised plant health advisory content from CABI, specifically processed and structured for Large Language Models (LLMs) and retrieval augmented generation (RAG) applications in agricultural advisory services.

The dataset contains expert-curated content from across CABI's international development and publishing business, covering topics such as integrated pest management, sustainable farming practices, and safe use of plant protection products.

This is a subset of CABI’s plant health content, with a focus on recently-published content relevant to the following countries:

Kenya
Ethiopia
India

For further content, please contact CABI directly at [email protected].

With a total of 476K tokens, this corpus provides a detailed knowledge base, enabling advanced AI models to generate accurate, localised responses to support informed decision-making in agriculture. Whether for automated knowledge retrieval, chatbot development, or scientific analysis, this dataset serves as a robust foundation for AI-driven advisory in the agricultural domain.

Each document has been systematically processed from its original JATS 1.3 XML format into Markdown (MD format) while preserving critical scientific context, metadata, and domain-specific agricultural knowledge. Rather than applying chunking or token-based segmentation, the CABI workflow is designed to retain the complete document structure in a single file, ensuring that the logical flow of the article—including headings, sections, tables, and references—is preserved end-to-end.

During processing, a standardized YAML front matter block is generated for every Markdown file. This front matter captures essential metadata such as:

publication identifiers
article type
authors and affiliations
geographic information
pest and crop information
subject descriptors and domain tags

This guarantees machine-readability, reproducibility, and interoperability with knowledge retrieval and LLM pipelines.

Dataset Contents

Dataset	Documents	Format	Description
Pest Management Decision Guides (PMDG)	272	Markdown + YAML metadata	Integrated Pest Management (IPM) advice designed to support agriculture extension officers with decision making for prevention, monitoring and control of specific country–crop–pest combinations, co-written with local experts. Developed through the PlantwisePlus program https://www.cabi.org/plantwiseplus/
Plantwise Factsheets For Farmers (PFFF)	49	Markdown + YAML metadata	IPM advice and step-by-step instructions on how to carry out specific pest management interventions, designed for agriculture extension officers and co-written with local experts. Developed through the PlantwisePlus program https://www.cabi.org/plantwiseplus/
Plant Health Cases	25	Markdown + YAML metadata	Case studies of plant health in practice, designed for students, lecturers, researchers and research-led practitioners

Total documents: 346
Total tokens: ≈ 476K

Ethical Notes

This dataset contains expert agricultural guidance.

It must not be used to conduct experiments that cause harm to human subjects.
AI systems built from it should consult local regulations and agronomists.
Pesticide information contained within it has been checked against local regulations and registrations at the point of writing. Any products or content derived from it should also be subject to checks against local pesticide regulations and registrations in order to prevent harm to humans, animals, and the environment.

Licence

Released under CC BY-NC 4.0, meaning reuse is permitted as long as attribution to CABI is maintained and it is used for non-commercial purposes only.

If you would like to use this dataset for commercial purposes, please contact CABI directly at [email protected].

Terms of Use

In addition to the licence listed above, any User of this dataset must comply with the following terms.

Restrictions on the use of this dataset:

The User will not use this dataset, or individual works therein, in any way which is not permitted under applicable law. In particular, the User will:

Not produce outputs that contain the whole or substantial parts of the dataset or individual works therein, verbatim;
Not sell, distribute, or publicly display this dataset, or individual works therein, themselves. Any such downstream circulation must be done by referring to this dataset and attributing the same to CABI using the Citation provided below;
Not use the dataset to reverse engineer or reconstruct proprietary elements of the same, or individual works within it;
Only process any personal data within the dataset, or in any individual work therein, in accordance with locally applicable data protection law(s). Where such data protection laws are not present, the UK GDPR will be deemed as the applicable data protection law, and any personal data must be processed in accordance with the same.

Security

The User will handle this dataset, or individual works therein, in accordance with good industry practice, including storage in a secure environment.

The User will inform CABI of any data breach within 24 hours of it occurring, or when the User becomes aware of such breach, whichever is sooner, especially of any personal data that is part of this dataset, or in individual works therein, and will inform CABI of the steps taken to mitigate harm from such breach, including any steps required to be taken in accordance with applicable law.

Attribution

The User must provide attribution to CABI as the content provider in any display or output resulting from or derived from this dataset or any individual work therin.

Attribution must at minimum include the Citation as provided below.
The Citation must be made visible and easily accessible by placing it on the main website of the User, preferably under a section that covers patrons, funders or partners.
The User must inform CABI in writing if it intends to remove the attribution and may only do so after receiving an explicit written waiver from CABI. In no circumstances can attribution by removed or deemed as waived unilaterally by the User.
The User acknowledges that attribution does not imply any endorsement by CABI of the User's activities with respect to this dataset or any individual work therin and must not expressly or implicitly misrepresent such endorsement.

Sublicences

The User may only give access to this dataset or individual work therin to sublicensees provided they comply with these terms and conditions and any other access requirements as are imposed by CABI. The User will be responsible for ensuring this compliance by the sublicensee(s).

Intellectual Property Rights

Between the User and CABI, CABI owns all intellectual property rights in this dataset.

Representations and warranties

The User warrants and undertakes to CABI that:

It will comply with all applicable laws and regulations, in performing its obligations and exercising its rights of use of this dataset, in the design, development, and deployment of any AI solution;
Any outputs and displays generated, displayed or otherwise made available by an AI solution will comply with all applicable laws and regulations, including laws relevant to data protection;
Any outputs from an AI solution will not infringe the intellectual property rights or other proprietary rights of any third party;
The AI solution will not produce discriminatory, harmful, or unethical outputs or displays and that the User will implement reasonable measures to identify and mitigate bias, misinformation and harmful content in such outputs and displays; and
The outputs and displays will not falsely imply any endorsement by or affiliation with CABI, unless CABI expressly approves in writing.

Dispute resolution

Any dispute or claim regarding this dataset or its use in accordance with these terms and conditions shall be exclusively adjudicated and/or settled in the courts of England and Wales, which shall have exclusive jurisdiction in this regard.

Citation

If you use this dataset, please cite:

CABI. Plant Health Content dataset. Source: Hugging Face (Online). https://huggingface.co/datasets/CABInternational/Plant-Health-Content [date accessed]

Acknowledgements

This dataset was developed for the Generative AI for Agriculture (GAIA) project funded by the Gates Foundation and the Foreign, Commonwealth and Development Office (FCDO).

Maintainers

Organization: CABI
Email: [email protected]

Downloads last month: 10