A knowledge-based scenario framework to support intelligent planning characters

Agent technologies have been successfully applied to craft believable characters in interactive scenarios. However, intelligent characters specialized for a controlled scenario with a predefined story are inadequate for open-ended scenarios. The key to deal with the open-endedness problem lies in the characters' ability to understand and analyze unexpected situations. Thus, an explicit representation of the environment is crucial. We found ontologies in the form of interconnected concepts to be an appropriate high-level representation because it enables character agents to reason about the world through inference. This paper proposes a knowledge-based framework for the construction of agent-based scenarios. The physical properties of the environment are dynamically converted to instances of concepts. We also show how an intelligent planning character, without any implicit knowledge about the scenario, can exploit the resources in the environment to make plans. With ontology support, characters show better adaptability and utilize the environment more creatively.