Rules View

Using the Knowledge Builder API - Introduction

Basic Inference from Cases

The inference process works by firing the rules based logic representation of the cases table, where each line in the table is considered as a rule. The order of firing is left to right (i.e. by attribute column) and then top to bottom (i.e. by each case). Note that you can just drag and drop columns to change the attribute order, and do the same with case lines by using drag and drop on the line numbers.

It is important to appreciate that more than one case (i.e. multiple rules) can fire when using Case inference. This means that all the attributes must be evaluated. The presence of "*" in values provides a shorthand for entering a case for each of the Attribute's values.

The presence of an Unassigned outcome in Cases table at runtime would result in no value being assignment to the Cases object (Decision). If however the Cases lead more than one Decisions at runtime and the Object is not set to be Multi-select, then the a fatal Clash runtime error would occur.

• If the isDynamic Object Control Property is set to False then the Object's 'Cases' inference will only be executed the first time it is encountered by the inference.
• The Object Control Property for knowledgeMode can be set to control if runtime inference for this object is use the Cases or not (e.g. from Decision Tress if one exists).
• The OnCapture & OnLeave events (where they enabled by the Tree Object's Base Class) are executed before the start and after the end of the tree inference respectively.
• Cases inference can follow different strategies (e.g. Constraint Inference, Case-based Inference, Fuzzy Inference etc) See Advanced Inference Techniques.