Suicidal Ideation Detection on Social Media: A Machine Learning Approach

Abstract

Mental illness is a huge problem among the population, identifying the individual who is at risk of suicide is necessary and is the first and foremost step to averting suicide. The purpose of the study is to distinguish between suicidal and non-suicidal posts that have been gathered from social media. The pre-processed data was utilised to perform a variety of machine learning algorithms, including Support Vector Machine, Logistic Regression, and AdaBoost, as well as term frequency-inverse document frequency for embedding. The results indicate that Support Vector Machine has the highest precision (80.72%), while Logistic Regression has the best accuracy (80.75%) and recall (77.81%).