Common Decision Patterns
IN OUR TOP-DOWN JOURNEY FROM high-level business requirements to decision service implementation, we have now reached the basement. Like most basements, it contains a slightly random collection of items. My aim is to present some concrete examples of how decision-making is automated in a few specific situations, to provide some solid ground for the rather abstract principles discussed in the previous chapters, and to provide some tricks of the trade for those new to BRMS. First, I present some examples of common types of automated decision, then I close with a discussion of various ways to handle collaborative decision-making where humans and decision services need to cooperate to reach a decision.
They say there's no substitute for experience, and after you have conducted a few knowledge automation projects you see the same decision patterns cropping up again and again with minor variations. Recognizing these patterns saves a considerable amount of time in the analysis process and allows the use of a toolkit of standard implementation types for development, as discussed in Chapter 5. The simplest patterns are what we could call “primitives,” consisting of just a single knowledge area in a standard format. These primitives include rule sets and metaphors, algorithms, and analytic models, as discussed in Chapter 3. Obviously all of these should be available as standard implementation types. But your toolkit should go further ...