The Large Language Model-based Emotional Counseling System for Intervening in Negative Emotions Among College Students

Objective: To develop a psychological counseling system based on large language models, and to compare its effectiveness in addressing issues such as depression, anxiety, and stress among college students, compared with various general-purpose large models. Methods: The system was developed through fine-tuning of psychological corpus data, design of structured prompts, multi-dimensional human evaluation, and proper system architecture. A baseline model without any prompts was used as a control group. A mixed design was employed, with four groups: the psychological counseling system, the baseline model, “DouBao,” and “DeepSeek.” This design involved two weeks of intervention followed by one week of follow-up for 235 college students. Results…