Moreover, we propose a novel prompt tuning based on the contrastive learning module that reformulates the downstream fake news detection task in a similar form as the graph contrastive pretraining, ...