Identification of College Students' Depressive Tendency Through Text Big Data Analysis

Abstract

Accurate identification of depressive tendencies in the early stage is beneficial to the treatment and prevention of depression. This paper presents a text-based depressive tendency recognition algorithm used to assist the judgment of depressive tendencies, which employs an intelligent algorithm to classify and identify texts and determines whether the user who publishes the text information is a patient with depressive tendency. The long short-term memory (LSTM) algorithm was used to recognize text information, and the convolution operation in the convolutional neural network (CNN) was introduced to improve the recognition performance of LSTM. After that, the convergence of the algorithm and the impact of various text vectorization on the algorithm were examined using Chinese texts on the Sina Weibo platform in the simulation experiment, and the results were compared with the support vector machine and traditional LSTM algorithms. It was found that the LSTM algorithm improved by using a CNN not only trained faster but also performed better in identifying depressive tendencies compared to the other two algorithms. The contribution of this paper lies in utilizing a CNN to further extract textual features, thereby enhancing the recognition performance of LSTM and providing an effective reference for accurately and quickly identifying depressive tendencies among college students.