| Popis: |
Agent-based social simulation has been comprehensively applied in the research of social and ecological systems. At its core is an artificial population, which endogenously drives the system evolution for particular applications, such as urban transportation, reginal economics, analysis of infectious disease transmission, and military simulation. In contrast with the previous population simulations where simple mathematical models are used to ‘reproduce’ actual demographic features, this paper proposes a self-evolutionary digital population system, named as Cognitive Artificial Population System (CAPS). At a more fine-grained level, CAPS focuses on the agent cognitive, reasoning and learning process in their surrounding environment, thus can exploit most advantages from cognitive computing and Artificial Intelligence. As a case study, Chinese population evolution is implemented using the proposed framework. Computational experiments indicate that CAPS is able to achieve good predicted population structures for real social systems. |
