Paper Title
EMOJI AND SENTIMENT-BASED SUICIDE IDEATION DETECTION MODEL

Abstract
According to the World Health Organization, more than700,000 people die due to suicide every year. It also the fourth leading cause of death among young people aged 15-29. Early detection of suicide ideation is important for effective suicide prevention. In contemporary times, individuals, especially young people, are more likely to share their thoughts on social media platforms. This report aims to develop a machine learning model based on emoji and sentiment analysis to detect suicide ideation in social media text. The data used for this study was collected from the popular social media platform Reddit, comprising a total of 232,074 texts. Various tools were employed for emoji and sentiment analysis to extract relevant features. Machine learning models were then used to train and test the data. Following the evaluation of the classification models, Support Vector Machine (SVM) was chosen with an accuracy of 92% and an F1-score of 92%. The model was then implemented in my data product to predict whether the input text reflects suicide-related content or non-suicidal text. Keywords - suicide ideation detection; emoji; sentiment; machine learning