Poster: A Social Media Pre-Training Framework for Rumor Detection Utilizing User Feedback

Social media platforms connect users with users around the world and allow everyone to post their opinions freely. This facilitates the spread of rumors. Identifying rumors on social media becomes a thorny challenge. Existing research has achieved good results by using the propagation characteristics of rumors. However, these researches only consider the impact of rumor propagation and ignore users' feedback on rumors. In this paper, we propose a bottom-up node aggregation method based on user feedback trees and look at how to use pre-training to model the structure of feed back trees.