AI/Ml Intervention Towards Detection and Prevention of Suicidality

Abstract

People's attitudes, emotions, and activities are the searchable archives on social media ( SM), thus providing an excellent opportunity to capture the behavioural attributes. The research on the possibility of leveraging Artificial intelligence-based models on social media is in its infancy. Our work focused on investigating the possibility of automatically detecting suicide- related posts on social media. The research started with the first objective of collecting a large dataset from two online platforms, Twitter and Reddit, to prepare the machine learning frameworks. After collecting the relevant data, the research problem is divided into two parts. One is to differentiate between suicidal and non-suicidal content. After data collection, human annotation was performed as per the proposed annotation scheme. There has been a detailed analysis of the methodology based upon the proposed advanced feature engineering mechanism, extracts and identifies the most relevant features, and then delivered to machine learning algorithms in order to expand accuracy.