Trust Risk Assessment Technology Based on Collaborative Analysis of Joint User Behavior

In the context of the current intensive and shared network environment, the passive security defense system faces challenges such as low security protection capabilities, difficulties in preventing internal threat attacks, and poor stability in system user behavior. To address these issues, this paper proposes a trust risk assessment technology based on collaborative analysis of joint user access behavior. By leveraging the deep learning algorithms of ConvLSTM and MLP, as well as the KNN machine learning algorithm, this approach enables real-time monitoring, collaborative analysis, and assessment of users' network and terminal behavioral deviations within a comprehensive spatio- temporal context. It further identifies potential security threats and anomalies, achieving multi-dimensional continuous security detection and malicious threat alerting within the system. Finally, an experimental optimization and evaluation of the model's threat detection capabilities were conducted. The results demonstrate that the prediction accuracy for terminal behavior exceeds 94%, while the prediction accuracy for network behavior surpasses 96%, providing robust supplementary support for safeguarding system security and ensuring efficient incident management.