Propagation of Information in a Semantic Net

In creating a computer system which can function as an ophthal-mological consultant, particular attention must be paid to the choice of a representation for the system's medical knowledge. This choice affects both the system's efficiency, and its ability to explain itself by reference to this knowledge. We describe a representation consisting of a semantic net in which a set of rules can be associated with any link. The system uses this representation to perform in a bottom-up, event-driven manner. Information propagates through the semantic net under control of the rules. We show how both efficiency and explanation are served by this representation.