The PROTEGE II project at Stanford KSL is probably the most well known such project. PROTEGE II [Eriksson et al. 1994] consists of a suite of tools that faciliate knowledge acquisition. MAITRE allows Knowledge Engineers to create domain ontologies, in effect the terminology of the domain. A tool called DASH takes as input a domain ontology and generates automatically a knowledge-acquisition tool that domain experts can use to populate the knowledge base with detailed content. The final step is to connect the acquired knowledge with a problem solving method by connecting the terminology of the domain to the requirements specfied by the problem solving method. The MARBLE tool accomplishes this goal. This approach has been used to construct problem solvers using a variety of problem solving methods in a variety of domains. The knowledge acquisition tools tend to be very stylized ``form filling'' type applications and this seems to work well enough in well constrained domains.
The EXPECT system at ISI has some of the same flavor, but a different technical approach. EXPECT [Swartout and Gil 1995] [Gil and Melz 1996] is based on the Explainable Expert System framework. Domain facts and problem solving principles are explicitly represented in two separate knowledge bases, each with relevant acquisition tools. A specific problem solver is produced by an Application Program Writer, in effect an automatic programmer which extracts the relevant facts and problem solving methods from these two knowledge bases and then generates the code for the specific problem solver. The design record maintained in this process makes the system highly capable of generating explanations of its behavior.
In both these systems, there has been an attempt to allow the domain experts to easily contribute to the construction of the system. This is a goal we share. However, these systems have tended to concentrate on limited technical domains rather than on acquiring a broad base of world knowledge. Our focus is on this more broad based information. Of course, if our system is to be used in one of these technical domains we would be interested in using whatever assets these systems have produced.