Classification tree models
To see an example click here.
A classification tree provides a hierarchical arrangement of concepts. This is an arrangement in which classes are shown to be made up of subclasses, which themselves are composed of sub-subclasses, and so on. Each node identifies a class name (in normal type) and the class's special features (in italics, using feature syntax). Classes automatically inherit the features of their ancestors.
At runtime, InterModeller attempts to match the user's data against one or more of the classes which occur at the leaf nodes of the classification tree. Note that non-leaf classes are never presented as advice. First, the program checks (by asking questions) that the feature descriptions of the root class are satisfied. If so, a subclass of the root is selected (the one that that has the uppermost node graphically) and its feature descriptions are checked in the same way. If some feature is not satisfied then an alternative subclass is investigated, otherwise the first subclass's subtree is searched exhaustively before any alternatives are considered. When a node at the end of some branch is reached with all features satisfied, InterModeller recommends the class name of that node. If the user then asks for more answers, InterModeller backtracks to investigate any unexplored alternatives.
Click below to learn more.