AI-augmented Cybersecurity Requirements Generation using LLMs | Reproducible Research Package

Fuente: Zenodo
Guardado en:
Detalles Bibliográficos
Autores principales: Yelmo, Juan Carlos, Martín, Yod-Samuel, Perez-Acuna, Santiago
Formato: Recurso digital
Lenguaje:inglés
Publicado: Zenodo 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866901167779348480
author Yelmo, Juan Carlos
Martín, Yod-Samuel
Perez-Acuna, Santiago
author_facet Yelmo, Juan Carlos
Martín, Yod-Samuel
Perez-Acuna, Santiago
contents <h1>AI-augmented Cybersecurity Requirements Generation using LLMs | Reproducible Research Package</h1> <p>This repository accompanies the paper “Experimental Evaluation of AI-Augmented Cybersecurity Requirements Generation Leveraging LLMs’ Capabilities” (<a href="https://doi.org/10.1109/ACCESS.2026.3658339">10.1109/ACCESS.2026.3658339</a>). It contains every script, dataset, prompt template and result needed to fully reproduce our empirical study.</p> <h2>Research Description</h2> <div> <div>This project investigates the practical use of state‑of‑the‑art Large Language Models (LLMs) to transform high‑level, standard‑driven cyber‑security controls into concrete, system‑specific requirements. Using a synthetic yet industrially plausible case study—AI4I4, an IoT‑enabled automotive logistics platform—we benchmark thirteen frontier models (GPT‑4, LLaMa 3, Mistral, QWen, etc.), representing tge state of the art as of September 2024, across four prompting pipelines and three temperature regimes.</div> <br> <div>Key contributions include:</div> <br> <div>1. <strong>Annotated benchmark</strong> of 54 ISO‑27002 clauses with placeholder semantics suitable for automatic instantiation.</div> <div>2. <strong>LangChain pipelines</strong> that decompose the task into applicability filtering, domain‑element search, requirement generation, and JSON formatting.</div> <div>3. <strong>Comprehensive evaluation</strong> of accuracy (precision, recall, F2), creativity (F2‑synthetic), and consistency (Jaccard overlap across runs).</div> <div>4. <strong>Prompt library</strong> enumerating >180 templates, showing how subtle changes in instruction design affect hallucination rate and coverage.</div> <br> <div>The artefacts and scripts below allow full replication—from raw prompts to final figures—on any infrastructure with access to the referenced models.</div> <div> </div> <div>For more information on the repository structure, reproducibility, licensing, and contact details, please refer to the README.</div> </div>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_17868180
institution Zenodo
language eng
publishDate 2025
publisher Zenodo
record_format zenodo
spellingShingle AI-augmented Cybersecurity Requirements Generation using LLMs | Reproducible Research Package
Yelmo, Juan Carlos
Martín, Yod-Samuel
Perez-Acuna, Santiago
LLM
Software Engineering
Requirements Engineering
Cybersecurity
Artificial Intelligence
<h1>AI-augmented Cybersecurity Requirements Generation using LLMs | Reproducible Research Package</h1> <p>This repository accompanies the paper “Experimental Evaluation of AI-Augmented Cybersecurity Requirements Generation Leveraging LLMs’ Capabilities” (<a href="https://doi.org/10.1109/ACCESS.2026.3658339">10.1109/ACCESS.2026.3658339</a>). It contains every script, dataset, prompt template and result needed to fully reproduce our empirical study.</p> <h2>Research Description</h2> <div> <div>This project investigates the practical use of state‑of‑the‑art Large Language Models (LLMs) to transform high‑level, standard‑driven cyber‑security controls into concrete, system‑specific requirements. Using a synthetic yet industrially plausible case study—AI4I4, an IoT‑enabled automotive logistics platform—we benchmark thirteen frontier models (GPT‑4, LLaMa 3, Mistral, QWen, etc.), representing tge state of the art as of September 2024, across four prompting pipelines and three temperature regimes.</div> <br> <div>Key contributions include:</div> <br> <div>1. <strong>Annotated benchmark</strong> of 54 ISO‑27002 clauses with placeholder semantics suitable for automatic instantiation.</div> <div>2. <strong>LangChain pipelines</strong> that decompose the task into applicability filtering, domain‑element search, requirement generation, and JSON formatting.</div> <div>3. <strong>Comprehensive evaluation</strong> of accuracy (precision, recall, F2), creativity (F2‑synthetic), and consistency (Jaccard overlap across runs).</div> <div>4. <strong>Prompt library</strong> enumerating >180 templates, showing how subtle changes in instruction design affect hallucination rate and coverage.</div> <br> <div>The artefacts and scripts below allow full replication—from raw prompts to final figures—on any infrastructure with access to the referenced models.</div> <div> </div> <div>For more information on the repository structure, reproducibility, licensing, and contact details, please refer to the README.</div> </div>
title AI-augmented Cybersecurity Requirements Generation using LLMs | Reproducible Research Package
topic LLM
Software Engineering
Requirements Engineering
Cybersecurity
Artificial Intelligence
url https://doi.org/10.5281/zenodo.17868180