Personalized Response System for Different Voice Phishing Types: Utilizing a Retrieval-Augmented Generation Model

This study presents the development of a comprehensive system designed to effectively classify voice phishing types and provide personalized response manuals tailored to specific scenarios. By integrating a binary classification model with a Retrieval-Augmented Generation (RAG) model, our system offers a dynamic solution to combat the evolving threats of voice phishing. The system's ability to generate situation-specific response manuals demonstrates a significant advancement in preventive security measures. Upon receiving voice phishing conversation data, the system classifies the type of phishing attempt and provides tailored response instructions based on the classified scenario. This approach highlights the system's potential as a practical tool for both enhancing security measures and providing educational resources.