Detecting Suicidal Tendencies through Machine Learning and Deep Learning Techniques

Detecting Suicidal Tendencies through Machine Learning and Deep Learning Techniques

B.Sindhuja, K.Shilpa, A.Satya Sindhu, K.Shiva Shankar vara prasad, Ms.Ila Chandana Kumari

Hyderabad Institute of Technology and Management (HITAM)

Abstract: This paper proposes a method to detect signs of suicide in individuals by analyzing their social media posts. The study focuses on publicly accessible Reddit data and employs machine learning algorithms and text representation techniques such as TF-IDF and Word2Vec to classify posts as either suicidal or non-suicidal based on their textual features. The models evaluated in this study are a Convolutional Neural Network (CNN) and a Random Forest model. The importance of early detection and prevention of suicide attempts is highlighted, and social media is identified as a potential platform to identify individuals at risk of suicide. The study demonstrates that the CNN model performs better than the Random Forest model in analyzing textual features to detect signs of suicide. The findings suggest that analyzing social media posts can be an effective approach to identifying individuals at risk of suicide and intervening before a suicide attempt occurs.

Keywords: Suicide prevention, social media, Text analysis, Machine learning, Convolutional Neural Network, Random Forest model.