多特征融合的可移植谣言早期检测模型

Rumor early detection model with multi-feature merged and portability

针对当前诸多网络平台的谣言泛滥现象,提出结合长短期记忆(Long-short Term Memory,LSTM)网络与支持向量机(Support Vector Machine,SVM)的可移植谣言早期检测模型。将谣言文本转换为向量序列,通过LSTM网络挖掘谣言文本的深层特征,并引入有效度、敏感度与热度特征。通过SVM融合训练拟合表明,该模型在多平台数据集上表现出良好的预测结果。

In view of the phenomenon of rumor overspreading among many platforms,a portable rumor early detection model with the combination of Long-short Term Memory(LSTM)network and Support Vector Machine(SVM)is proposed.Vector sequences converted from rumor corpus are fed into LSTM network to mine the hidden text feature.Effectiveness,sensitivity and heat features of rumor corpus are introduced and merged by SVM training.The experimental results show that the model performs well in multi-platform dataset.