ANALYSIS OF INTELLIGENT METHODS FOR SOLVING THE PROBLEM OF FUNCTIONAL INFORMATION SECURITY PROFILES FORMATION

Abstract

This paper explores the potential of applying various artificial intelligence methods to solving the problem of functional information security profile formation. Functional profiles serve as the foundational basis for secure automated systems creation, yet current formation methodologies are often hindered by subjectivity and a lack of repeatability due to heavy reliance on human expertise. The study aims to analyze intelligent means that can minimize the human factor.
The research provides a detailed comparative analysis of several key artificial intelligence technologies, including self-organizing maps, large language models, expert systems, genetic algorithms, and fuzzy logic. Each method is evaluated based on specific characteristics such as adaptability, interpretability of results, implementation complexity, and dependence on training data. Self-organizing maps are assessed for their clustering capabilities but are noted for their low interpretability. Large language models, particularly when integrated with retrieval-augmented generation, show high adaptability in providing justified recommendations based on cybersecurity standards. Expert systems are examined for their deterministic nature which ensures logical correctness, while genetic algorithms are considered for multi-objective optimization of security controls. Furthermore, fuzzy logic is highlighted for its capacity to handle uncertain linguistic variables and subjective expert assessments.
The findings indicate that while individual methods have specific strengths, no single method fully satisfies the requirements for transparency, accuracy, and regulatory compliance simultaneously. Consequently, the authors propose a hybrid conceptual approach that utilizes an expert system as the logical core to ensure strict adherence to regulatory requirements, integrated with a large language model that acts as an interpreter to transform formal rules into clear, human-readable justifications. This synergy significantly enhances the objectivity and efficiency of the information security profiling process by reducing expert bias.