Classification of the Suicide-Related Text Data Using Passive Aggressive Classifier

Abstract

AbstractAs technological advancements have been increasing in day-to-day life, most people are relying on Internet usage where they are using social media platforms to convey their status through their respective profiles. As social media made a big boom on the planet, most people use these applications to share their views in many ways like text, images or videos, etc., with which a huge amount of data gets generated. From the data related to their posts, it is been observed that most people are not able to get rid of their mental stress. As the report of WHO (World Health Organization) shows that suicides are the second-largest global pandemic, our objective is to analyze the text within the suicide notes posted on Twitter (one among the social media platforms). Text classification is been done on the data, and we have many machine learning algorithms, neural networks, regression techniques, etc., Like the large stream of data is been generated to get better results, we use passive aggressive classifier (PAC) algorithm. Along with PAC, we use SVM, Naive Bayes, random forest, and decision tree and have a comparative study. Since we considered the labeled textual dataset, we apply the above models on the dataset and classify and show the result as “Suicidal” or “Non-Suicidal.”KeywordsText classificationSVMRandom forestNaïve BayesDecision treeSklearnPassive aggressive classifierSuicide-related