PSYCHOLOGICAL STRESS PROBLEM SENSING METHOD AND DEVICE

摘要

The embodiments of the present application provide a psychological stress problem sensing method and device. Wherein, the above method comprises: obtaining a corresponding initial word vector based on text vocabulary and combining an LSTM network layer to obtain a context vector; obtaining a semantic association matrix based on an MHA model, and obtaining the semantic association degree mapping in combination with a fully connected network layer; generating an attention vector based on the semantic association degree mapping and an attention network layer; combining the context vector and a feedforward fully connected network model, to obtain a stress classification vector. The psychological stress problem sensing method and device provided by the embodiments of the present application use the LSTM network layer, the MHA model, the fully connected network layer, the attention network layer and the feedforward fully connected network model to obtain the stress classification vector, take a large amount of text vocabulary reflecting the psychological state of adolescents as the basis of data analysis, perform a large amount of data training by a computer to accurately obtain the sensing results reflecting the psychological stress state of the adolescents, which are accurate and effective, and provide a basis for psychological counseling of the adolescents.