摘要
为了解决传统谣言检测算法无法提取文本的深层语义信息及文本序列特征的问题,提出基于改进Transformer的双向编码器表示(Bidirectional Encoder Representation from Transformers,BERT)模型的谣言检测方法。该方法融合了BERT和长短期记忆网络(Long Short Term Memory,LSTM)模型的优点,可以有效对长文本谣言进行分类。实验结果表明,该方法的分类效果优于经典的谣言检测模型。
In order to solve the problem that traditional rumor detection algorithms can not extract the deep semantic information and text sequence features,a rumor detection method based on the improved Bidirectional Encoder Representation from Transformers(BERT)model is proposed.This method combines the advantages of Bert and Long Short Term Memory(LSTM)model,and can effectively classify long text rumors.Experimental results show that the classification effect of this method is better than that of the classical rumor detection model.
作者
韦金琼
WEI Jinqiong(School of Computer and Information Engineering,Nanning Normal University,Nanning Guangxi 530000,China)
出处
《信息与电脑》
2022年第15期1-3,共3页
Information & Computer
关键词
分类模型
谣言检测
文本分类
classification model
rumor detection
text classification