Repository accompanying the paper:
Making Environmental Models Understandable for Non-Experts: LLM-Based Explanations with a Peatland Case Study
This repository contains supplementary materials supporting the paper and is intended to improve transparency and reproducibility.
.
├── prompts/
│ └── raw_prompt_template.md
├── personas/
│ └── persona_evaluation_readme.md
├── expert_evaluation/
│ └── Peatland_Expert_Evaluation_Example.pdf
├── rag/
│ └── (grounding documents)
├── figures/
└── README.md
- Raw explanation-generation prompt template
- Persona-based evaluation framework
- Expert evaluation form and example
- Documentation describing the evaluation workflow
The proposed framework combines:
- Machine-learning prediction
- Retrieval-Augmented Generation (RAG)
- LLM-based explanation generation
- Multi-persona evaluation
- Iterative refinement
- Optional human expert feedback
The objective is to generate scientifically grounded, understandable explanations of peatland water-table predictions for non-expert stakeholders.
The raw prompt template used for explanation generation.
The complete persona-based evaluation procedure together with scoring criteria and example feedback.
An example evaluation document used by peatland experts.
The repository is provided to facilitate peer review and reproducibility.
To protect project assets, the following are not included:
- proprietary datasets
- trained ML models
- confidential project material
If you use these materials, please cite:
@misc{davoodi2026peatland,
title={Making Environmental Models Understandable for Non-Experts: LLM-Based Explanations with a Peatland Case Study},
author={Laleh Davoodi},
year={2026},
note={Supplementary materials repository}
}Laleh Davoodi
University of Turku
Email: ladavo@utu.fi
© 2026 Laleh Davoodi, University of Turku.