The emerging technologies such as Metaverse and large-scale pre-trained models are driving deep integrations among humans, as well as humans and machines. This introduces significant uncertainties in the management and control of complex cyber-physical-social systems, which encompass virtual spaces, physical systems, and human societies. Social computing, particularly human-centered social computing, is a viable, effective, and even the only approach to address such challenges. This chapter delves into the modeling and analysis of human users, as well as their behavior prescription, through a multifaceted approach that encompasses virtual population synthesis, individual cognitive decision-making, computational deliberative experiments, and exemplary case studies. By adopting this approach, the goal is to gain a deep understanding of human user demands and behaviors within complex cyber-physical-social systems. This, in turn, enables the development of a robust and reliable experimental methodology for managing and controlling large-scale, highly dynamic human-in-the-loop systems. Ultimately, it enables the development of human-centered AI systems that prioritize adaptability, reliability, and seamless user interaction.