Transformer-enhanced Hawkes process with decoupling training for information cascade prediction

and paths to simulate the coupled topological and temporal information for capturing the dependencies between the nodes in a graph. At the second-level, a pooling attentive module is proposed to embed the cascade evolution rate for modeling short-term outbreaks. Extensive experiments on two real-world datasets demonstrate the significant performance improvements of Hawkesformer over existing state-of-the-art models.