Deep Learning Approach for the Detection of Depression in Twitter

Abstract

Depression is a common and serious mental illness that negatively affects from all the ways like how you feel, the way you think and how you act. And also it has very complicated nature. These reasons makes it difficult to find out a better method for detecting mental illness such as depression among social media users. Recent research studies have shown that the popularity of online social media networks among one’s life is getting increased day by day. Also online social media users makes it as a good platform for expressing their emotions and feelings. Choosing appropriate supervised machine learning approaches for identifying mental illnesses through social media platforms (especially Twitter) are somewhat difficult due to the unavailability of proper amounts of annotated training data. So here try to create some sufficient amount of data and implement deep neural architecture for the same. A combination of Long Short Term Memory and Convolutional Neural Network, and Support Vector Machine (SVM) are used to detect users with signs of mental illnesses (depression in this case) from the Twitter social media platform. The evaluation results shows that Deep Neural Network is producing more accurate results compared to Machine Learning algorithm like SVM.