Psychological Counseling Cannot Be Achieved Overnight: Automated Psychological Counseling Through Multi-Session Conversations

In recent years, Large Language Models (LLMs) have made significant progress in automated psychological counseling. However, current research focuses on single-session counseling, which fails to capture the ongoing nature of real-world counseling processes. Effective counseling is not a one-time interaction but an ongoing process, where each session builds upon the last to progressively address a client's issues. To overcome this limitation, we introduce a dataset for Multi-Session Psychological Counseling Conversation Dataset (MusPsy-Dataset). Due to the sensitive nature of counseling and inherent privacy concerns, we construct this synthetic dataset from authoritative, publicly available psychological case reports. The MusPsy-Dataset is designed to reflect the temporal continuity and shifting mental states of a single client across multiple sessions. Leveraging this resource, we propose the MusPsy-Model, which tracks client progress and dynamically adjusts counseling goals over time. Experiments show that our model performs better than baseline models across multiple sessions.