Scholarlink

Abstract <p>Over the past decade, cyberbullying has become a pressing issue among youth, garnering significant societal attention. The advancement in detection methods has evolved from traditional single-mode data analysis to leveraging multimodal data fusion technology. This work presents an innovative model, multimodal dynamic propagation graph capsule network, aimed at detecting cyberbullying. By integrating a cross-modal co-attention mechanism, this work explores correlations across multimodal content within the original post. Furthermore, it meticulously captures dynamic interactive features during the cyberbullying propagation process through a dynamic propagation graph capsule network. Extensive experimental evaluations conducted&#xa0;on real-world datasets reveal that our framework surpasses&#xa0;current state-of-the-art models for detecting cyberbullying.</p>

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

Cyberbullying Detection Using Multimodal Dynamic Propagation Graph Capsule Network

  • 期刊论文

  • 发表日期：

    2026年3月1日

查看全文
  • Yongheng Chen,
  • Chunyan Yin

摘要

Abstract

Over the past decade, cyberbullying has become a pressing issue among youth, garnering significant societal attention. The advancement in detection methods has evolved from traditional single-mode data analysis to leveraging multimodal data fusion technology. This work presents an innovative model, multimodal dynamic propagation graph capsule network, aimed at detecting cyberbullying. By integrating a cross-modal co-attention mechanism, this work explores correlations across multimodal content within the original post. Furthermore, it meticulously captures dynamic interactive features during the cyberbullying propagation process through a dynamic propagation graph capsule network. Extensive experimental evaluations conducted on real-world datasets reveal that our framework surpasses current state-of-the-art models for detecting cyberbullying.