A0388
Title: Supervised centrality via sparse spatial autoregression
Authors: Yingying Ma - Beihang University (China) [presenting]
Wei Lan - Southwestern University of Finance and Economics (China)
Chenlei Leng - University of Warwick (United Kingdom)
Ting Li - Hong Kong Polytechnic University (Hong Kong)
Hansheng Wang - Peking University (China)
Abstract: The social characteristics of the players in a social network are closely associated with their network positions. Identifying the influential players in a network is of great importance as it helps to understand how ties are formed and how information is propagated, and in turn, can guide the dissemination of new information. A new notion of supervised centrality is proposed, emphasizing that the centrality of a player is task-specific. A novel sparse spatial autoregression is developed by introducing individual heterogeneity to each user to estimate the supervised centrality and identify important players. To overcome the computational difficulties in fitting the model for large social networks, a forward-addition algorithm is further developed, and it is shown that it can consistently identify a superset of the influential nodes. The method is applied to analyze three responses in Henan Floods data: the number of comments, reposts and likes, and obtain meaningful results. A simulation study further corroborates the developed theory.