Závěrečná práce: Bc. Matúš Maťaš: Automated Generation of Vulnerability Descriptions Using Local LLMs for Penetration Testing Reports
Diplomová práce
Automated Generation of Vulnerability Descriptions Using Local LLMs for Penetration Testing Reports
Anotace
Táto diplomová práca sa zaoberá návrhom a hodnotením systému na automatizované generovanie štruktúrovaných nálezov zraniteľností s využitím Large Language Models (LLMs). Teoretická časť opisuje úlohu takýchto modelov v kybernetickej bezpečnosti a metódy ich adaptácie. Praktická časť sa zaoberá prípravou trénovacej sady odvodenej z reálnych výsledkov penetračných testov, implementáciou dvoch odlišných …více
Abstract
This master’s thesis addresses the design and evaluation of a system for automated generation of structured vulnerability findings in pen- etration testing reports leveraging Large Language Models (LLMs). The theoretical part describes the role of such models in cybersecurity, methods for their adaptation including prompt engineering, retrieval- augmented generation and fine-tuning, and the challenges …více
Zadání práce
This thesis will focus on the design and implementation of a tool for automated retrieval and summarization of vulnerability information based on CVE or similar identifiers. Using locally hosted large language models, the system collects data from public sources (e.g., NVD) and anonymized penetration testing reports provided by an industry partner. It then generates structured vulnerability descriptions in English, including key fields such as impact and remediation. The thesis addresses problem analysis, data preprocessing, implementation of locally hosted LLM and prototype development that integrates into existing reporting workflows. The tool’s performance will be evaluated against historical penetration test reports and assessed by security professionals for accuracy, clarity, and language quality.
25. 5. 2026 15:56, Ing. Katarína Galanská, učo 422705
Konzultant
Citace dle normy ČSN ISO 690
Práce na příbuzné téma
Seznam prací, které mají shodná klíčová slova.
-
Formulace statistických příkladů a projektů v éře generativní umělé inteligence
Bc. Lenka Šoltésová -
Predicting Facebook Ad Campaign Performance Using Few-Shot Learning with Large Language Models
Mgr. Shahadat Hussain -
Enhancing Knowledge Retrieval with Language Models
Bc. Michal Cikatricis -
Scalable spatio-temporal analytics framework for urban networks
Jura Bogomolov, Ph.D. -
Anotace analýzy sentimentu a postoje pomocí generativního jazykového modelu
Mgr. Daniel Šurda -
LLMs with Test Feedback for Program Synthesis
Bc. Marcel Nadzam -
Bias Analysis in Medical Exam Questions Generated by Large Language Models
Mgr. Ema Madarászová -
Automation of the Request for Quotation Process using Large Language Models
Mgr. Martin Kňažovič, učo 553635




