ABSTRACT
In building ever more sophisticated expert systems, knowledge engineers are plagued by the bottleneck o f needing to insert more and more knowledge into the machine. One long-range solution is for the program-by itself-to learn via discovering. The first case study presented, AM, demonstrates that new domains of knowledge can be developed mechanically by using heuristics. Yet as new domain concepts, facts, and conjectures emerge, specific new heuristics-infor mal judgmental rules-are needed. They in turn can be discovered by using a body o f heuristics for guidance. The second case study, E u r i s k o , has already achieved some promising results in this endeavor. To achieve reasonable perfor mance it has proven useful to hypothesize at least the rudiments of a theory of heuristics. Work with E u r i s k o indicates that, as new domains o f knowledge emerge and evolve, new (at least augmented) representations for knowledge are needed. We believe that these can be developed by using heuristics, and this chapter concludes by presenting an example o f E u r i s k o devising new, useful slots.
