Detecting Suicidal Ideations in Online Forums with Textual and Psycholinguistic Features DOI Creative Commons
Eldar Yeskuatov, Sook-Ling Chua, Lee Kien Foo

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(21), P. 9911 - 9911

Published: Oct. 29, 2024

Suicide is a global public health problem that takes hundreds of thousands lives each year. The key to effective suicide prevention early detection suicidal ideations and timely intervention. However, several factors hinder traditional risk screening methods. Primarily, the social stigma associated with presents challenge ideation detection, as existing methods require patients explicitly communicate their propensities. In contrast, progressively more at-risk people choose online platforms—such Reddit—as preferred avenues for sharing experiences seeking emotional support. As result, these platforms have become an unobtrusive source user-generated textual data can be used detect suicidality supervised machine learning natural language processing techniques. this paper, we proposed approach combines psycholinguistic features extracted from Reddit forum. Subsequently, selected most informative using Boruta algorithm employed four classifiers: logistic regression, naïve Bayes, support vector machines, random forest. Bayes models trained combination term frequency-inverse document frequency (TF-IDF) National Research Council (NRC) demonstrated highest performance, obtaining F1 score 70.99%. Our experimental results illustrate yields better classification performance compared those separately.

Language: Английский

Detecting Suicidal Ideations in Online Forums with Textual and Psycholinguistic Features DOI Creative Commons
Eldar Yeskuatov, Sook-Ling Chua, Lee Kien Foo

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(21), P. 9911 - 9911

Published: Oct. 29, 2024

Suicide is a global public health problem that takes hundreds of thousands lives each year. The key to effective suicide prevention early detection suicidal ideations and timely intervention. However, several factors hinder traditional risk screening methods. Primarily, the social stigma associated with presents challenge ideation detection, as existing methods require patients explicitly communicate their propensities. In contrast, progressively more at-risk people choose online platforms—such Reddit—as preferred avenues for sharing experiences seeking emotional support. As result, these platforms have become an unobtrusive source user-generated textual data can be used detect suicidality supervised machine learning natural language processing techniques. this paper, we proposed approach combines psycholinguistic features extracted from Reddit forum. Subsequently, selected most informative using Boruta algorithm employed four classifiers: logistic regression, naïve Bayes, support vector machines, random forest. Bayes models trained combination term frequency-inverse document frequency (TF-IDF) National Research Council (NRC) demonstrated highest performance, obtaining F1 score 70.99%. Our experimental results illustrate yields better classification performance compared those separately.

Language: Английский

Citations

0