Scholarlink

<p>Recently, the number of fake news has been continuously increasing, and its concealment has been constantly enhanced, which poses a significant threat to social stability and public cognition. Therefore, achieving early detection of fake news is particularly important. However, existing methods often struggle to effectively capture the multi-scale temporal and structural features involved in the news dissemination process and the unique temporal dynamics of the dissemination process, thereby limiting the improvement of detection performance. To address this issue, we propose a novel early fake news detection model based on hypergraph—MTSHG. This model can comprehensively describe the complex interaction patterns among multi-dimensional entities such as users, news content, and social relationships during the news dissemination process by decomposing the hypergraph into multiple subgraphs. Additionally, we design a multi-level temporal and structural attention mechanism to extract dynamic information at different time scales and multi-level structural characteristics during the learning process. The experimental results show that our MTSHG model increases the accuracy of fake news detection by 2.46% to 4.77%, 0.72% to 0.93%, 0.46% to 1.49%, and 0.68% to 2.81% in the four real datasets PolitiFact, GossipCop, Pheme, and Weibo, and also demonstrates high accuracy in early detection. Our code is publicly available at <a href="https://github.com/kobeimissyou/MTSHG">https://github.com/kobeimissyou/MTSHG</a>.</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

Early detection of fake news through integrating hypergraph-based multi-scale temporal and structural information

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Chao Liu,
  • Jiayi Chen

摘要

Recently, the number of fake news has been continuously increasing, and its concealment has been constantly enhanced, which poses a significant threat to social stability and public cognition. Therefore, achieving early detection of fake news is particularly important. However, existing methods often struggle to effectively capture the multi-scale temporal and structural features involved in the news dissemination process and the unique temporal dynamics of the dissemination process, thereby limiting the improvement of detection performance. To address this issue, we propose a novel early fake news detection model based on hypergraph—MTSHG. This model can comprehensively describe the complex interaction patterns among multi-dimensional entities such as users, news content, and social relationships during the news dissemination process by decomposing the hypergraph into multiple subgraphs. Additionally, we design a multi-level temporal and structural attention mechanism to extract dynamic information at different time scales and multi-level structural characteristics during the learning process. The experimental results show that our MTSHG model increases the accuracy of fake news detection by 2.46% to 4.77%, 0.72% to 0.93%, 0.46% to 1.49%, and 0.68% to 2.81% in the four real datasets PolitiFact, GossipCop, Pheme, and Weibo, and also demonstrates high accuracy in early detection. Our code is publicly available at https://github.com/kobeimissyou/MTSHG.