Towards popularity prediction of information cascades via degree distribution and deep neural networks

Xiaodong Feng, Qihang Zhao, Rui-Jie Zhu

Abstract: Highlights•A graph-level deep learning model is proposed for information cascade prediction.•It uses the degree distribution vector to represent the structure of the cascade.•The dynamic representation of the entire cascade is learned by sequential DL models.•It dramatically reduces the prediction error and costs much less time.