User Modeling in Social Media and Big Databy Qiang Yang
Professor, Department of Computer Science and Engineering
Hong Kong University of Science and Technology
The ever-growing social networks and social media provide invaluable sources of information for modeling the behavior of users. High-quality user models enable superior services and functions for end users. In this talk, I will present several examples of user modeling based on social networks and social media. I will describe our research in modeling users' information preferences on Microblogs using a novel user message model and discuss our work on extracting users' daily activities, such as dining and shopping, that inherently reflect their habits, intents and preferences. I explain our novel transfer learning solution via a collaborative boosting framework comprising a text-to-activity classifier for socially connected users. I will also describe our research on user modeling in multiple, overlapping social networks in a “composite social network” setting. Finally, I will explain our research on finding social spammers in large social networks.