A Framework for Identifying Excessive Sadness in Students through Twitter and Facebook in the Philippines

Abstract

Natural Language Processing (NLP) can be used to identify a person's sentiments or emotions. Depression is one sentiment that researchers have tried to identify through Natural Language Processing with little success. Depression is an episode of sadness or apathy, along with other symptoms, that lasts for at least two consecutive weeks. Depression is especially bad with students due to the amount of stress and anxiety they have to go through. While depression is very difficult to identify and treat, excessive sadness, one of the symptoms that may lead to depression can be identified early and appropriate action can be taken. The Philippines is known to have the highest depression count in Southeast Asia. Data Mining was performed on Twitter and Facebook, and with the use of Natural Language Processing (NLP) and Sentiment Analysis, a logistics regression model was devised with the use of emotion Lexicons to identify the user's state. The Latent Dirichlet Allocation (LDA) was then used to identify important topics of each user and cluster the data and make sense out of each user's excessive sadness.