Rag Classifier for proactive Cyberattack Diagnostics

As part of the research, a classification model based on the Retrieval- Augmented Generation (RAG) approach was developed and trained to proactively prevent threats and vulnerabilities in critical information infrastructure of government and corporate information systems. For this purpose, methods of systematic, comparative, logical and structural analysis, deep learning methods, including the RAG approach, were used. The model is trained on a semantic dataset of cyber threat descriptions, consisting of 3936 records and containing various text descriptions of several classes of events, including various types of cyber attacks (RAG Classifier). A key feature of the model is its ability to classify cyber threats based on a semantic description of a situation occurring in a critical information infrastructure entered by the user. The proposed model can be integrated into a single information security system for the critical information infrastructure of government and corporate information systems.