Classification of Depression using Temporal Text Analysis in Social Network Messages


Gabriel Melo, KaykeBonafé and Guilherme Wachs-Lopes, University Center of FEI, Brazil


In recent years, depression has gained increasing attention. As with other disorders, early detection of depression is an essential area of study, since severe depression can result in suicide. Thus, this study develops, implements, and analyzes a computational model based on natural language processing to identify the depression tendencies of Twitter users over time based on their tweets. Consequently, an F-measure of 83.58 % was achieved by analyzing both the textual content and the emotion of the papers. With these data, it is possible to determine whether constant fluctuation of emotions or the message in the text is a more accurate indicator of depression.


Depression, Natural Language Processing, Machine Learning.

Full Text  Volume 12, Number 15