Understanding the propagation of toxicity on social media platforms is necessary for ensuring safer digital spaces. This paper aims to measure the accuracy of the five most established epidemiological models on the spread of toxicity in Reddit. By applying these models to two datasets—one from a highly toxic subreddit that was banned and one from various groups throughout Reddit—we address a gap in the literature by evaluating their performance on Reddit data. The results indicate that epidemiological modeling can achieve a fitting error of less than 2% on toxicity spread in Reddit. Furthermore, we concluded that more sophisticated models such as SEIRS and SIRS performed significantly better than simpler models like SIS or SIR. This research successfully demonstrates the effectiveness of epidemiological modeling on Reddit and paves the way for further research and potential prevention strategies for toxicity on Reddit.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Modeling Toxicity Propagation on Reddit Using Epidemiology

  • Selimhan Dagtas,
  • Nitin Agarwal,
  • Niloofar Yousefi

摘要

Understanding the propagation of toxicity on social media platforms is necessary for ensuring safer digital spaces. This paper aims to measure the accuracy of the five most established epidemiological models on the spread of toxicity in Reddit. By applying these models to two datasets—one from a highly toxic subreddit that was banned and one from various groups throughout Reddit—we address a gap in the literature by evaluating their performance on Reddit data. The results indicate that epidemiological modeling can achieve a fitting error of less than 2% on toxicity spread in Reddit. Furthermore, we concluded that more sophisticated models such as SEIRS and SIRS performed significantly better than simpler models like SIS or SIR. This research successfully demonstrates the effectiveness of epidemiological modeling on Reddit and paves the way for further research and potential prevention strategies for toxicity on Reddit.