Loading...
Paper Type
Complete
Abstract
Toxic content, including abusive and hateful conversations, is a growing concern on social media platforms. This paper proposed and evaluated a novel approach based on generational parent comments and a tree structure to detect and predict toxic comments and toxic triggers in online dialogues on Reddit using nine machine learning algorithms. Specifically, we study the influence of generational parent comments on the toxicity of their child comments study how toxicity permeates them. Our approach achieves high accuracy in predicting the toxicity of Reddit comments. We discover that the immediate parent comment has the most influence on the toxicity of a comment. The research also showed toxicity tends to stop online discourse. This research enhances understanding of social media toxicity, aiding policymakers (Government officials) and social media moderators in early detection and prevention.
Paper Number
1865
Recommended Citation
Falade, Tope Christopher Christopher; Yousefi, Niloofar; and Agarwal, Nitin, "Toxicity Prediction in Reddit" (2024). AMCIS 2024 Proceedings. 18.
https://aisel.aisnet.org/amcis2024/social_comp/social_comput/18
Toxicity Prediction in Reddit
Toxic content, including abusive and hateful conversations, is a growing concern on social media platforms. This paper proposed and evaluated a novel approach based on generational parent comments and a tree structure to detect and predict toxic comments and toxic triggers in online dialogues on Reddit using nine machine learning algorithms. Specifically, we study the influence of generational parent comments on the toxicity of their child comments study how toxicity permeates them. Our approach achieves high accuracy in predicting the toxicity of Reddit comments. We discover that the immediate parent comment has the most influence on the toxicity of a comment. The research also showed toxicity tends to stop online discourse. This research enhances understanding of social media toxicity, aiding policymakers (Government officials) and social media moderators in early detection and prevention.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SOCCOMP