Empathy-Enhanced Chatbot for Psychological Support: A Retrieval-Augmented and Therapy-Informed Approach

In modern society, busy work schedules and environmental factors contribute to high levels of stress, negatively impacting mental well-being. This issue is especially critical for people who often experience prolonged stress with limited access to psychological support. This study proposes a chatbot system designed to provide real-time, empathetic, and effective psychological assistance. Unlike conventional chatbots that offer superficial responses, our system integrates multiple psychotherapy techniques to deliver deeper emotional resonance and sustained support. Built on a large language model (LLM), the chatbot is enhanced with Retrieval-Augmented Generation (RAG) and optimized for personalized, context-aware interactions. The system adopts RA-CoP, a novel framework that combines RAG with structured psychotherapeutic guidance, as its core mechanism for generating emotionally attuned and clinically informed responses. This study not only proposes the RA-CoP framework and designs a psychological support system, but also introduces an evaluation method that combines objective scoring with users' subjective experiences to assess the usability of such systems.