一种基于深度学习模型的常识类谣言自动检测方法

An Automatic Detection Method of Common Sense Rumors Based on Deep Learning Model

近年来通过网络爆发的重大舆情事件数量激增,谣言自动检测成为净化网络生态环境的重要研究课题的大背景下,本文主要探讨利用LSTM+Attention深度学习技术对常识类谣言进行自动检测的解决方案,与传统的CNN、MLP、SVM等模型进行对比,并通过实验证明所提出的模型能捕捉到常识类谣言的特征,在常识类谣言的检测任务上比传统的机器学习算法提升超过10%的精度。

In recent years,the number of major public opinion incidents that have erupted through the Internet has increased sharply.Under the background that automatic rumor detection has become an important research topic for purifying the network ecological environment,this article mainly discusses the solution of automatic detection of common sense rumors using LSTM+Attention deep learning technology,it is compared with traditional CNN,MLP,SVM and other models,and experiments prove that the proposed model can capture the characteristics of common sense rumors,and the accuracy is improved by more than 10%compared with traditional machine learning algorithms in the detection task of common sense rumors.