Arabic Fake News Detection Using Deep Learning

Document Type

Article

Publication Date

Winter 1-1-2024

Abstract

In recent years, the explosive growth of social media platforms has led to the rapid spread ofvast amounts of false news and rumors on the internet. This disrupts various news sources such as onlinenews outlets, radio and television stations, and newspapers, especially in Arab countries. Therefore, the fakenews detection problem has been raised worldwide. Arabic research in this field is very little comparedto English research. Previous researchers had used machine learning and deep learning techniques on alarge scale, but recently they used pre-trained models in their studies. Our proposed model works by usingthe Arabic pre-trained Bidirectional Encoder Representations from Transformers (Arabic BERT) to extractfeatures from the text, then uses a Convolutional Neural Network (1D-CNN or 2D-CNN) to reduce the sizeof the features and extract the important ones, then passes it to an artificial neural network to perform theclassification process. In our experiment we introduce a novel hybrid system consists of two main parts.In the first part we try three Arabic pre-trained Bidirectional Encoder Representations from Transformersmodel (APBTM) which are AraBERT [1], GigaBERT [2] or MARBERT [3], while in the second part, we use1D-CNN or 2D-CNN. this leads to six architectures from this system. we make our experiment by train andevaluating every architecture using three datasets which are ( Arabic News Stance (ANS) [4], AraNews [5],and Covid19Fakes [6]). A comparison is made between the proposed model and other modern models whichused the same dataset. We made three sets of experiments depending on the used datasets. Each set includesa group of experiments, and then we present the results in tables. Our proposed model which is the hybridmodel between AraBERT and 2D-CNN has achieved the best F1-scores of 0.6188,0.7837 and 0.8009 whenusing the ANS dataset, the Ara-News dataset, and the Covid19Fakes dataset respectively. Furthermore, themodel reduces the training time by achieving better results with less number of epochs. The results indicatethat the proposed model offers the best performance, with 71% accuracy in the Arabic News Stance (ANS)[4] dataset outperforming the model made by Sorour et al. [7] and the model made by W.Shishah et al. [8]that achieved accuracy of 67% and 66% respectively (PDF) Arabic Fake News Detection Using Deep Learning.

Link to Full Text

Share

COinS