A Survey on Prediction of Suicidal Ideation Using Machine and Ensemble Learning. (17th November 2019)

• Record Type: Journal Article
• Authors: Chadha, Akshma; Kaushik, Baijnath
• Is Part Of: Computer journal. Volume 64:Number 11(2021)
• Page Start: 1617
• Page End: 1632
• Publication Date: 2019-11-17
• Subjects: suicidal ideation -- twitter -- multinomial naïve bayes -- bernoulli naïve bayes -- decision tree -- logistic regression -- support vector machine -- random forest -- AdaBoost -- voting ensemble; Computers -- Periodicals; 005.1
• DOI: 10.1093/comjnl/bxz120
• Languages: English
• ISSNs: 0010-4620

Abstract

Abstract: Suicide is a major health issue nowadays and has become one of the highest reason for deaths. There are many negative emotions like anxiety, depression, stress that can lead to suicide. By identifying the individuals having suicidal ideation beforehand, the risk of them completing suicide can be reduced. Social media is increasingly becoming a powerful platform where people around the world are sharing emotions and thoughts. Moreover, this platform in some way is working as a catalyst for invoking and inciting the suicidal ideation. The objective of this proposal is to use social media as a tool that can aid in preventing the same. Data is collected from Twitter, a social networking site using some features that are related to suicidal ideation. The tweets are preprocessed as per the semantics of the identified features and then it is converted into probabilistic values so that it will be suitably used by machine learning and ensemble learning algorithms. Different machine learning algorithms like Bernoulli Naïve Bayes, Multinomial Naïve Bayes, Decision Tree, Logistic Regression, Support Vector Machine were applied on the data to predict and identify trends of suicidal ideation. Further the proposed work is evaluated with some ensemble approaches like Random Forest, AdaBoost, Voting Ensemble to see the improvement.