CyberHelp: Sentiment Analysis on Social Media Data Using Deep Belief Network to Predict Suicidal Ideation of Students

Abstract

Suicide is a very critical and important issue in modern society. Suicide is the third-leading cause of death for college and high school students. Social media allows students in the digital environment to share their suicidal ideas and thoughts with others. Accurate and early detection and prevention of suicidal ideation in students can save the students' lives. To identify the risk factor for suicidal attempts, a suitable method of analysing the suicidal behaviour of students using their sentiment text posted on social media can be used. This paper presents an optimized Dragonfly algorithm (DFA) using a Deep Belief Network (DBN) for the automatic detection of suicidal ideation in students. In our CyberHelp Solution, the proposed DFA-based DBN model analyses student social media data, predicts suicidal behavior, and treats students appropriately. The sentiment analysis performs automated categorization of online messages and makes accurate predictions of the student’s suicidal behaviors. The dragonfly heuristic optimization algorithm is used for tuning the hyperparameter in the deep belief network. The proposed DFA-DBN technique has been implemented to predict suicidal ideation in students with a higher accuracy of 95.5% compared with other classification models.