Artificial society generation for modern video games

Video games have long had a need for realistic non-player characters or agents driven by some form of artificial intelligence. Recently, researchers and developers have spent considerable effort towards creating more believable or humanlike agents by borrowing concepts from the social sciences. Agents that exist and function only in isolation, however, lack the connectedness typically associated with believability, and so there is a need for broader social context and community for agents. This paper presents an approach to generating a society of believable agents with human-like attributes and social connections. This approach allows agents to form various kinds of relationships with other agents in the society, and provides a basic form of shared or influenced attributes based on familial relationships. Our proposed method provides a solid foundation for artificial society generation, and a prototype implementation of this approach shows great potential for future work. As a modularized and parameterized framework, there are also many opportunities for extending the system or customizing it to the needs and requirements of a particular game or application.