Depressive Emotion Recognition Based on Behavioral Data

  With the increase of pressure in people’s lives, depression has become one of the most common mental illness worldwide. The wide use of social media provides a new platform for depression recognition based on people’s behavioral data. This study utilizes the linguistical psychological characteristics of Weibo users to predict users’ depression level. The model adopts the Gaussian process regression algorithm, sets the PUK kernel as the kernel function, applies the forward-backward search method to select feature, and uses five-fold cross-validation to evaluate performance of the model. This study finally established a prediction model with a correlation coefficient of 0.5189, which achieved a medium correlation in the psychological definition, and provided a more accurate method for the auxiliary diagnosis of depression.