Abstract

AbstractSocial media, with the characteristics of fast spread and easy access, has become a main channel for people to obtain news. However, in this situation, fake news spread faster and farther, which has a negative impact on national politics, economy and culture. Therefore, how to automatically detect fake news from numerous news has become significantly important. Most of the current detection algorithms mainly explore and extract features from text characteristics of news, obtaining less effective, because the fake news on social media does not exist independently. More auxiliary information is needed to enhance the accuracy of detection. In this paper, a new fake news detection model Dual Graph Attention Networks (DGAT) is proposed, which utilize three entities (news, publishers and users), existing in social media platforms, and their relationships to construct heterogeneous information networks. DGAT model consists of two layers of graph attention networks. The node level attention network is responsible for learning the different weights of the same-type neighbor nodes. The type level attention network is responsible for learning the different weights of different types. The experimental results show that DGAT model outperforms all baseline models in terms of accuracy, precision, recall and F1 value on FakeNewsNet datasets.KeywordsFake news detectionHeterogeneous Information NetworksGraph attention networksSocial media

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call