Academic Team Influence Spreading Prediction Based on Community Representation Learning

The overall influence of the academic team is determined by the influences of all individuals and the internal structure of the team. The influence spreading network can reflect the influence of individual, but it lacks the structural information of the team. Therefore, we propose to integrate the academic cooperation network into the influence spreading network to improve the forecast accuracy of team influence spreading. In this paper, the deep learning is employed to get community vector representation of the team. We take the degree of node in academic cooperation network as the importance of team members, and sort them. Community representation is generated by RNN-based sequential autoencoder, and we combine it with the feature representation of the node in influence spreading network to predict team influence spreading. Experiments on real data show the effectiveness of community representation learning and the academic team influence spreading prediction scheme.