Researchers

Ahmad Hussein, Nada Ghneim and Ammar Joukhadar

Published in

Damascus University Journal for Engineering Sciences, First conference on Informatics, volume 38, issue 4, October 2022.

Abstract

Recently, online social networks, like Twitter, Facebook, Instagram, and others have revolutionized interpersonal communication and allowed millions of users of different ages and genders to develop their social and professional relationships, which increased spreading false information and fake news. Fake news is especially prevalent in the events and pandemics like the covid-19 pandemic, leading to individuals accepting bogus and potentially deleterious claims and articles. Quick detection of fake news can reduce the spread of panic and confusion among the public. In this article, we present our

approach to analyze the credibility of Arabic information on social media, which is presented in the form of a two-step pipeline. The first step classifies the tweet if it contains information or not, and the second step calculates the distance between the tweet text and the titles obtained from the search results to calculate the credibility of the tweet.

We built an Arabic annotated data set of 5,000 tweets. The proposed approach was evaluated on built dataset and on NLP4IF 2021 dataset. The results showed that the results on the built dataset were better and it equals 0.91.

Keywords: Credibility, Social Media, High Impact Events, Fake News, Neural Networks, Transfer Learning.

Link to full paper

http://journal.damascusuniversity.edu.sy/index.php/engj/article/view/6395/1577