Graph Neural Network for Fake News Detection and Classification of Unlabelled Nodes at MediaEval 2022

In this paper we describe our approach to fake news detection for the MediaEval 2022 challenge that has run for the third time. As in the previous editions, the goal of the challenge is the detection of misinformation tweets, but in this edition, both text and graph data are provided. We focus on the classification of unlabelled nodes/users in the graph by utilizing graph neural networks to classify them as either fake news spreader or just an ordinary node i.e. non fake news spreader. Apart from those labels, the classification apply for unlabelled nodes in conspiracy theories related to COVID-19 in nine different categories. Furthermore, graph based node classification detection for whole categories will be done since this will lead to more comprehensive classification analysis rather than just to label them either as a spreader or non spreader of fake news.

Download paper here

Bibtex

@article{maulana2022graph,
  title={Graph Neural Network for Fake News Detection and Classification of Unlabelled Nodes at MediaEval 2022},
  author={Maulana, Asep and Pogorelov, Konstantin and Schroeder, Daniel Thilo and Langguth, Johannes},
  year={2022}
}