Rumor Detection Based on Attention CNN and Time Series of Context Information

Yun Peng,Jianmei Wang

doi:10.3390/fi13110267

Abstract

This study aims to explore the time series context and sentiment polarity features of rumors’ life cycles, and how to use them to optimize the CNN model parameters and improve the classification effect. The proposed model is a convolutional neural network embedded with an attention mechanism of sentiment polarity and time series information. Firstly, the whole life cycle of rumors is divided into 20 groups by the time series algorithm and each group of texts is trained by Doc2Vec to obtain the text vector. Secondly, the SVM algorithm is used to obtain the sentiment polarity features of each group. Lastly, the CNN model with the spatial attention mechanism is used to obtain the rumors’ classification. The experiment results show that the proposed model introduced with features of time series and sentiment polarity is very effective for rumor detection, and can greatly reduce the number of iterations for model training as well. The accuracy, precision, recall and F1 of the attention CNN are better than the latest benchmark model.

Highlights

Introduction and Time Series of ContextBy the end of 2020, China had 989 million netizens and 70.4% Internet penetration, with more than 1.6 billion mobile Internet users
Divided the whole rumor recognition model into three modules: Capture, Score and Integrate, in which the Capture module uses RNN to learn the time representation of the text. Their experimental results are better than previous work (SVM-time series (TS) [18], SVMcom DTS [19]) which prove the effectiveness of deep neural network model in rumor detection
Compared with the DT-Rank algorithm, the accuracy of the LK-Radial Basic Function (RBF) model is increased by 3.3%, the accuracy is increased by 3.7% and the recall is improved by 4.0%

Summary

Introduction

Introduction and Time Series of ContextBy the end of 2020, China had 989 million netizens and 70.4% Internet penetration, with more than 1.6 billion mobile Internet users. With the popularity of the Internet, microblog, WeChat and other applications gradually occupy people’s lives. They become important platforms for publishing and collecting information. Taking Weibo as an example, on the one hand, Sina Weibo provides sentiment outlets for the public. It forms a hotbed for making and disseminating rumors [1]. Rumors on social media are rampant and it is difficult to distinguish between credible and untrustworthy information, which can lead to social unrest and seriously endanger national security. It is important to detect rumors in the early development of rumors

Objectives

Methods

Results