Keraia: A Knowledge Engineering and Reference AI Architecture

The original intent of artificial intelligence was unveiled at the Dartmouth AI Summer Research Project in 1956. During a 2-month study, it was conjectured that every aspect of learning or any other feature of intelligence could in principle be so precisely described that a machine could be made to simulate it. The question asked then was how one could make machines use language, form abstractions and concepts, and solve the kinds of problems now reserved for humans. The question remains relevant today. This paper presents a reference AI architecture, dubbed Keraia, a symbolic knowledge engineering platform that implements the original intent. The platform includes a repository that serves three purposes: (1) a repository for captured human expertise; (2) an execution platform where previous solutions to known problems can be reused and applied to exact or similar problems; and (3) a collaboration platform where a society of domain experts can create new solutions or modify previous solutions. In support of these objectives, four core features—a novel knowledge representation language, a reference architecture, a general-purpose paradigm builder and a knowledge engineering methodology were developed. The target is to provide domain experts with tools to model precisely how they see their world and how they solve problems. As a case study, we adopt an implementation of the risk board game. It requires expertise and, in these knowledge-rich environments, we introduce a practical knowledge acquisition process and a platform supporting that process.