Machine Learning-Driven Analysis of Suicidal Language on Social Media: A Dive into Natural Language Processing (NLP) Techniques

Suicide is a serious worldwide problem affecting people of different ages. An alarming trend has already started, where individuals share their negative thoughts on various social media platforms which leads them to depression and gradually to self-harm. Social media posts can be a good indicator of a person's mental state. So, it is important to identify those negative posts to rescue people from this kind of unwanted problem. Our research highlights that using Machine Learning (ML) and Natural Language Processing (NLP) methods can help find people at risk and make it easier to take quick actions that might save lives. In this paper, We have used a dataset representing different emotions and situations discussed in the social media posts. Here we have used different types of ML algorithms including Naive Bayes, Support Vector Machine (SVM), Random Forest, Decision Tree, XG-Boost, and Gradient Boost to classify the posts as suicidal and non-suicidal. Remarkably, the SVM algorithm emerged as the most effective, achieving an accuracy of 92.67 % and an f1-scorc of 0.93. In contrast, the Gradient Boost algorithm yielded the lowest performance.