Predicting life satisfaction based on the emotion words in self-statement texts.

Abstract

Measuring people's life satisfaction in real time on a large scale is quite valuable for monitoring and promoting public mental health; however, the traditional questionnaire method cannot fully meet this need. This study utilized the emotion words in self-statement texts to train machine learning predictive models to identify an individual's life satisfaction. The SVR model was found to have the best performance, with the correlation between predicted scores and self-reported questionnaire scores achieving 0.42 and the split-half reliability achieving 0.939. This result demonstrates the possibility of identifying life satisfaction through emotional expressions and provides a method to measure the public's life satisfaction online. The word categories selected through the modeling process were happy (PA), sorrow (NB), boredom (NE), reproach (NN), glad (MH), aversion (ME), and N (negation + positive), which reveal the specific emotions in self-expression relevant to life satisfaction.