User-Characteristic Enhanced Model for Fake News Detection in Social Media

In recent years, social media has become an ideal channel for news consumption while it also contributes to the rapid dissemination of fake news out of easy access and low cost. Fake news has detrimental effects both on the society and individuals. Nowadays, fake news detection in social media has been widely explored. While most previous works focus on different network analysis, user profiles of individuals in the news-user network are proven to be useful yet ignored when analyzing the network structure. Therefore, in this paper, we aim to utilize user attributes to discover potential user connections in the friendship network with attributed network representation learning and reconstruct the news-user network to enhance the embeddings of news and users in the news propagation network, which effectively identify those users who tend to spread fake news. Finally, we propose a unified framework to learn news content and news-user network features respectively. Experimental results on two real-world datasets demonstrate the effectiveness of our proposed approach, which achieves the state-of-the-art performance.