Research on Psychological Decompression VR Scene Recommendation Algorithm Based on Graph Contrast Learning*
Zhuo Yang, Zhengping Li, Lijun Wang, Yuwen Hao, Xiaoxue Li · 2024
The present study investigates the potential of an emotion-based psychological decompression recommendation algorithm that utilises the SAM scale to assess users’ emotional states and employs a graph neural network to recommend personalised VR scenes. In doing so, the algorithm addresses the limitations of traditional stress relief methods that rely on in-person consultations with mental health professionals. Additionally, the study reviews current virtual reality emotion regulation techniques and their shortcomings, emphasising the significance of recommendation systems in mitigating information overload. The proposed algorithm introduces a novel interaction mechanism between users and objects by utilising user emotions, which replaces traditional behavioural signals. The Kolmogorov-Arnold Network (KAN) is employed to capture the nonlinear features of users and items, thereby enhancing the performance of the recommendation algorithm. Psychological experiments demonstrate that the user-vr-emotion interaction based on user emotions is effective in the graph comparison learning recommendation algorithm, resulting in more accurate VR scene recommendations for psychological decompression. This study delves into emotional classification methods, psychological decompression technology, and experimental design and data processing to validate the efficacy of VR scenes for psychological decompression. Comparative experiments confirmed the effectiveness and superiority of the proposed algorithm, offering new insights for future emotion-based personalised recommendation systems.