Scholarlink

This research introduces a novel approach to track symbolic (visual elements) transmission through digital networks by implementing disease propagation models from epidemiology. Focusing on the anti-disinformation campaign surrounding the 2024 Taiwan Presidential Election, the study analyzed 1953 YouTube videos from January 13th to 17th. The methodology integrated large language model-based symbol detection with epidemiological modeling and parameter optimization using Nelder-Mead and L-BFGS-B algorithms. Five epidemiological models (SIS, SIR, SIRS, SEIR, SEIZ) were evaluated for symbolic propagation patterns, with the SEIZ model proving most effective, suggesting symbols create lasting cognitive imprints that sustain transmission even after initial exposure wanes. The approach treats symbolic content as transmissible elements to measure spread rates and identify which visual elements propagate most effectively through information networks. This interdisciplinary methodology provides fresh perspectives on the mathematical modeling of information diffusion in digital spaces, with potential applications extending beyond disinformation to marketing virality and the propagation of cultural memes.

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 the Propagation Dynamics of Visual Elements with Epidemiological Frameworks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sayantan Bhattacharya,
  • Nitin Agarwal

摘要

This research introduces a novel approach to track symbolic (visual elements) transmission through digital networks by implementing disease propagation models from epidemiology. Focusing on the anti-disinformation campaign surrounding the 2024 Taiwan Presidential Election, the study analyzed 1953 YouTube videos from January 13th to 17th. The methodology integrated large language model-based symbol detection with epidemiological modeling and parameter optimization using Nelder-Mead and L-BFGS-B algorithms. Five epidemiological models (SIS, SIR, SIRS, SEIR, SEIZ) were evaluated for symbolic propagation patterns, with the SEIZ model proving most effective, suggesting symbols create lasting cognitive imprints that sustain transmission even after initial exposure wanes. The approach treats symbolic content as transmissible elements to measure spread rates and identify which visual elements propagate most effectively through information networks. This interdisciplinary methodology provides fresh perspectives on the mathematical modeling of information diffusion in digital spaces, with potential applications extending beyond disinformation to marketing virality and the propagation of cultural memes.