Rising Star Forecasting Based on Social Network Analysis

With the development of cyber-physical-social networks, researches on evaluating social influence have drawn increasing interests. Social influence indicates the importance of people in social networks. As a typical type of social networks, how to evaluate scholar influence in the academic social network has been a practical issue for research institutions. In this paper, we aim at evaluating the latent influence of scholars to find academic rising stars, which refer to scholars that may have few papers and little impact currently, but he or she will become influential scholars in the future. Most of the current works focus on the assessment of rising stars. However, there exists a growth period for each scholar. It is unfair for young scholars with limited resources, who will make acquaintance at conferences and recommendations and who will learn from senior scholars. In this paper, we primarily propose StarRank, which is an improved PageRank method to calculate the initial values of rising stars, construct the social network via explicit and implicit links, and apply the neural network to predict scholars' rankings in the future. The experimental results on real data set demonstrate that our method has a better performance than the-state-of-the-art methods on the count of hitting rising stars and the spearman correlation coefficient.