Finding a Depressive Twitter User by Analyzing Time Series Tweets

Abstract

Mental Health Status is a significant feature in human life, and it is observed through one's actions and behavior. This behavior is reflected in the changes in emotions, feelings, and loss of interest in previously enjoyed activities. Several researchers attempted to interpret social website data to find new events. Twitter is one of the prominent social sites with more than 330 million users sending a considerable number of tweets a day. We collected 0.2 million tweets related to the keywords of the Kessler Ten-point questionnaire and analyzed them. The tweet data set of a day is compared with our corpus to find the abnormality. We attempted to grade these depression keyword tweet time-series into four categories to isolate the anomaly. Our studies further suggested a process to identify a user from the tweets of a time series zone.