计算机应用研究2025，Vol.42Issue(10)：3012-3018,7.DOI:10.19734/j.issn.1001-3695.2025.03.0061
融合动态传播网络与双重特征差异的社交网络谣言检测模型
Rumor detection model in social networks integrating dynamic propagation networks and dual feature differences
摘要
Abstract
Most of the existing rumor detection models rely heavily on static information,making it difficult to reflect the dyna-mic propagation characteristics of rumors.Moreover,these models often overlook the feature differences and dynamic evolution in dimensions such as emotional polarity and thematic semantics between the original posts and comments.To solve this problem,this paper innovatively proposed an integrating dynamic network and dual feature differences model(DNDF)for rumor detection in social networks.The model aimed to improve detection effectiveness by analyzing the evolution of multidimensional features.Firstly,it used a dual feature difference module to analyze feature variations in emotional polarity and thematic semantics be-tween original posts and comment sequences.Then it combined propagation graph sequences and applied BiLSTM to generate dif-ferential feature sequences.Finally,it introduced a co-attention mechanism to strengthen the interactive learning between text features and emotional difference features,as well as between thematic difference features.Experiments on public datasets PHEME and WEIBO show that the DNDF model increases the accuracy by 0.3％and 2％respectively.The model outperforms mainstream baseline models in multiple indicators,such as F1,and confirms its effectiveness in rumor detection in social networks.关键词
特征差异/动态传播网络/谣言检测/社交网络Key words
feature difference/dynamic propagation network/rumor detection/social network分类
计算机与自动化引用本文复制引用
徐桂琼,何思桦,李卫民..融合动态传播网络与双重特征差异的社交网络谣言检测模型[J].计算机应用研究,2025,42(10):3012-3018,7.基金项目
国家社会科学基金资助项目(23BGL270) （23BGL270）
