Towards a unified AI formalism

This paper presents a unified approach to building intelligent architectures. Our approach relies on making some semantic, ontological, os well os architectural commitments. Semantically, we commit ourselves to principles governing the nature of the entities represented by the knowledge representation formalism, and the relationships between the faculties of reasoning, acting, and natural language understanding. More specifically, we concentrate on the nature and representation of entities, beliefs, actions, plans, and reasoning and acting rules. Based on these principles we present the design of an integrated AI architecture that has a unified knowledge representational formalism os well as a unified reasoning and acting component. The design of the architecture is based on object-oriented principles. We also show that such a design is amenable to implementation in a concurrent object-oriented programming paradigm. We demonstrate the advantages of our approach using several examples from our work.<>