Aquinas is a tool for creating knowledge bases for several different expert system shells. In this system, knowledge is represented by a set of rating grids. A rating grid consists of a number of problem solutions, which are differentiated by assigning weighted ranking values to their properties. Rating grids provide the basic data for performing “reasoning.” They are structured hierarchically within the knowledge base, grouping together problem solutions at equivalent levels of abstraction. The hierarchical structure also permits subgrouping of expert knowledge sources within problem-solving contexts. A knowledge base may therefore be assembled from modules created by a number of experts.
In the construction phase, a dialogue manager provides guidance to the expert. Each module can be tested and checked for inconsistencies before inclusion. Despite the interactive dialogue manager, Aquinas requires mastery of many advanced conceptual skills, and it would seem unlikely that a subject expert would be able to use Aquinas without assistance from a knowledge engineer. This is obviously important because both ease of use and time spent are directly related to the cost of knowledge base production.
The rating grid principle itself is not new. It can successfully cope with problems whose solutions can be enumerated in a satisfactory way, but I would suggest that only essentially trivial problems are amenable to this approach, not those that would demand consultation with an expert.
No reference is made to the appearance of the system to an end-user making inquiries of the knowledge base. If the system is to be used in environments outside the laboratory, it must be simple to understand and use. It must also be able to present sufficient background information to an inquirer, in a variety of formats, to allow reasoned judgments to be made.