Knowledge acquisition in the small: building knowledge-acquisition tools from pieces

The knowledge-systems community is interested in easing the knowledge-system development process. One approach, the mechanisms approach, views knowledge systems as a set of tasks, each of which can be realized by a computation mechanism. To be effective, knowledge-acquisition (KA) tools must be automatically configured once a set of mechanisms has been selected. We present a method for automatically generating a model-based KA tool for a given set of mechanisms. The method advocates combining KA mechanisms, which acquire knowledge in the small, and a set of strategies that provide a global view of the KA activity. We show that these global strategies are necessary for the KA tool to efficiently interact with a domain expert.