构建多平台谣言传播数据集，提升跨平台谣言识别能力
MPPFND: A Dataset and Analysis of Detecting Fake News with Multi-Platform Propagation
- 构建多平台谣言传播数据集MPPFND，捕捉跨平台传播结构
- 跨平台传播特征差异显著，影响谣言检测效果
- 基于图神经网络的APSL模型有效提升检测性能
虚假新闻在社交媒体中广泛传播，造成诸多负面影响。现有检测算法多聚焦于新闻内容与社交上下文分析，但通常仅针对特定平台，忽略了不同平台传播特征的差异。本文提出MPPFND数据集，涵盖多平台传播结构，并分析各平台的评论与传播特性，揭示其社交上下文存在显著差异。我们设计了一种多平台虚假新闻检测模型APSL，利用图神经网络提取多平台社会上下文特征。实验表明，考虑跨平台传播差异可显著提升虚假新闻检测性能。
原文摘要 · Abstract (English)
Fake news spreads widely on social media, leading to numerous negative effects. Most existing detection algorithms focus on analyzing news content and social context to detect fake news. However, these approaches typically detect fake news based on specific platforms, ignoring differences in propagation characteristics across platforms. In this paper, we introduce the MPPFND dataset, which captures propagation structures across multiple platforms. We also describe the commenting and propagation characteristics of different platforms to show that their social contexts have distinct features. We propose a multi-platform fake news detection model (APSL) that uses graph neural networks to extract social context features from various platforms. Experiments show that accounting for cross-platform propagation differences improves fake news detection performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。