2019-07-26 | 李哲鹏：社交关系网络中的热点预测
In social networks, social foci refer to physical or virtual entities around which social individuals organize joint activities, for example, places and products in physical form or opinions and services in virtual form. Forecasting which social foci will diffuse to more social individuals is important to business and public functions such as planning, marketing, and operations. Considering diffusive social adoptions, prior studies on user adoption behavior in social network contexts has focused on single-item adoption in homogeneous social networks. We advance this body of research by modeling the scenarios of multi-item adoption and learning for the relative spread of concurrent social diffusions of social foci in online social networking platforms. To be specific, we distinguish two types of social nodes, social foci and social actors by proposing a two-mode social network model. Based on social network theories, we identify and operationalize factors that drive social adoption, within the two-mode social network. We also capture the interdependences between social actors and social foci using a bilateral recursive process, namely, mutual reinforcement process that converges to an analytical form. Given the proposed model and the converged process, we thereby develop a gradient learning method based on mutual reinforcement process (GLMR) that targets on optimal parameter configuration for pairwise ranking of social diffusion spreads. Further, we show analytical properties of our method such as guaranteed convergence and convergence rate. In the evaluation, we benchmark against prevalent methods and show the superior performance of our method using three real-world data sets that cover adoptions about both physical and virtual entities from online social networking platforms.
Zhepeng (Lionel) Li is an Associate Professor in the Area of Operations Management and Information Systems at the Schulich School of Business, York University, Toronto, Canada. He has received his Ph.D. in Information Systems with a minor in computer science from the University of Utah, USA. His research broadly falls in computational data science for deep business analytics. Specific research interests concentrate on applying machine learning approaches to address business problems including social recommendations, targeted marketing, network analytics, FinTech, and PropTech. His works are published by top-tier research journals, such as Management Science and Information Systems Research. The published researches are also covered by media, such as MIT technology review. His research activities are supported by Natural Sciences and Engineering Research Council of Canada (NSERC) discovery grants and industrial sponsors.