An overview of knowledge manipulation engines and the intelligent file store

Reviews the hardware design requirements of knowledge manipulation engines (KME), and shows how one prototype KME attempts to satisfy these requirements. Knowledge manipulation engines are targetted at supporting large knowledge-based systems. The AI community is far from agreed upon what constitutes the best knowledge representation formalism or best front-end language for knowledge-based systems. KMEs therefore need to have a flexible approach to the choice of formalisms and computational models to which they give hardware support. Declarative languages are prime candidates for support. A KME should offer an architectural framework which allows `memory-based´ and `processor-based´ functionality to cooperate. The author gives the principal design requirements of the memory and processing activities of a KME