EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Rule Relaxation with Large Language Models in Agricultural Regulatory Technologies for Compliance and Permit Assessments

Ananda Maiti, Jinying Li

Computers · 25.10.2026 · 15 (11) · 755

DOI: 10.3390/computers15110755 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computers, online erschienen am 25.10.2026. Verfasst von Ananda Maiti und 1 weiteren Autor:innen. Die Arbeit nennt 60 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Rules form the foundation of regulatory systems by defining the obligations, permissions, and prohibitions that govern human activities. Regulatory processes universally rely on taking specific inputs from the stakeholders before checking them against specific rules. As regulators in various sectors such as agriculture increasingly adopt digital regulatory technologies, regulatory rules must be interpreted consistently by both humans and machines. Among the diverse rule types in legislation, quantitative rules are particularly important because they define measurable constraints that, if modified, may alter the scope of regulatory compliance. This paper presents a large language model (LLM)-based framework for analyzing rule relaxation in quantitative agricultural regulations. The proposed approach classifies quantitative rules into three categories: single-value rules, ranged or interval rules, and ratio rules. A mathematical framework is introduced to formally represent quantitative rule relaxation, while controlled numerical perturbations are applied to determine whether modifications to regulatory parameters expand, reduce, or ambiguously affect the set of actions permitted by a regulation. This methodology addresses the semantic effects of quantitative changes on deontic regulatory rules while preserving their underlying obligations and permissions. The framework is evaluated using pesticide regulations published by the Australian Pesticides and Veterinary Medicines Authority (APVMA). Performance is assessed using both class- and rule-specific accuracy and macro F1 scores for multiple LLM models. The results demonstrate that LLMs can be used to interpret the semantic consequences of quantitative modifications. However, the accuracy of performance declines with the number of variables and the complexity of the rules. The proposed framework establishes a practical foundation for explainable agricultural Regulatory Technology (RegTech) systems and future Rules as Code implementations.

Zitieren

APA

Maiti, A., & Li, J. (2026). Rule Relaxation with Large Language Models in Agricultural Regulatory Technologies for Compliance and Permit Assessments. Computers, 15(11), 755. https://doi.org/10.3390/computers15110755

Vancouver

Maiti A, Li J. Rule Relaxation with Large Language Models in Agricultural Regulatory Technologies for Compliance and Permit Assessments. Computers. 2026;15(11):755. doi:10.3390/computers15110755

BibTeX
@article{Maiti2026,
  author = {Maiti, Ananda and Li, Jinying},
  title = {Rule Relaxation with Large Language Models in Agricultural Regulatory Technologies for Compliance and Permit Assessments},
  journal = {Computers},
  year = {2026},
  volume = {15},
  number = {11},
  pages = {755},
  doi = {10.3390/computers15110755}
}
Kostenlos starten →