A brief description of each class follows:

Represents entire classification hierarchy. Has a name and description. Holds the root node of the classification hierarchy. Knows how to read and write itself from and to a file.

Corresponds to a concept or hypothesis in the problem domain. Embeds domain and control knowledge. Has a name, hypothesis, structured matcher, and interactions with other specialists. Knows how to perform establish-refine procedure. Knows how to establish its hypothesis. Knows its parent and child nodes.

Assists specialist in establishing or rejecting its hypothesis. Contains domain knowledge. Guides user through series of questions to capture case description. Produces confidence value expressing plausibility of hypothesis.

Supports abductor by capturing interactions between specialists. One specialist may expect or imply another's hypothesis. Has a strength that is either added or subtracted from the other specialist's confidence value depending on whether or not the hypothesis is expected.

Encapsulates information about a specialist, its confidence value, and its interactions for presentation to the abductor.

Produces a "best explanation" of a case description by selecting the subset of plausible hypotheses that best explains the case. Has a limit which influences how hypotheses are accepted or rejected.

Represents a problem instance to be evaluated by a classifier. For example, if a classifier diagnoses automotive problems a case is created for each specific car considered by the system. Has a name and symptoms (a textual description). Holds on to a classifier that is used to evaluate the case. Knows how to evaluate itself. Knows how to write and read itself to and from disk.