The goal of this research is to develop a prototype of a highly interactive "identification workbench" in the domain of plant- parasitic nematode genera. The system will be based on principles central to both expert systems and to expert workstations. There are important interactive systems questions, as the group of users is a diverse one, varying from novice to expert on domain knowledge, as well as familiarity with modern computing environments. Dealing with this diverse group will be handled through the "set of available tools" paradiam from expert workstations. Graphical assistance will be available to the identifier, and the object- oriented environment of Smalltalk presents a powerful and fully integrated environment in which objects will help unify the system model and the user's model of the identification process. The problem area is challenging and appropriate from the point of view of expert systems because nematode identification is highly problematic with weighty economic consequences, there are very few expert taxonomists and poor geographic distribution of existing ones, there is large variability within taxa, several important factors introduce different kinds of uncertainty, and the overall identification process is based on rules of thumb. Efforts to codify the knowledge in the domain and assist with the development of an underlying architecture are proceeding in parallel with this project. The concepts of protomorphs and nuclei will be fully developed and refined by a panel of about thirty of the approximately forty expert nematologist worldwide. These experts will also help characterize the identification process and evaluate the finished prototype.
