This article was authored by Francis Ip

What is a rule? One of the dictionary definitions is “A generalized statement that describes what is true in most or all cases: In this office, hard work is the rule, not the exception.” In essence, a rule governs the permissible behaviour within a given context and set of conditions. Using this definition as baseline, let us explore the way in which we classify rules.

Dichotomy of Rules

We can divide rules into two broad categories; namely, natural and man-made.

Examples of Natural Rules:

• When a person throws an object up in the air at sea level, it comes back to earth most of the time. Nevertheless, when the initial velocity of the object is at approximately 40,200 km/h (or 25,000 mi/h), the object may never come back down to earth.
• Pure water is in solid form when the ambient temperature is at or below zero degrees Celsius under normal atmospheric pressure. It turns into liquid form when the temperature is above zero and below 100 degrees Celsius. It changes into vapour form when the temperature is above 100 degrees Celsius.
• When one flips a coin, it lands either head or tail. It may stand on its edge in extremely rare occasions according to Catastrophe Theory.
• When one rolls a dice, one of the possible six face values turns up.

The first two examples are deterministic, while the last two are probabilistic. When applying the Law of Large Numbers from Statistics, we can make inferences about probabilistic rules. In the long run, head shows up 50% of the time and face value 6 shows up one sixth of the time.

Examples of Man-made Rules:

• Every vehicle must stop when a traffic light signal is “red”. A vehicle may make a right turn when it is safe to do so in some provinces in Canada and some states in the US. [Traffic Rules]
• No employee can use computers at work for his or her personal affairs. [Computer Usage Rules]
• When an authorization for purchase plus current credit balance exceeds a set credit limit, increase the credit card holder’s credit limit automatically in multiples of $100, if the credit card holder has a good credit rating and a consistent payment history. [Customer Credit Rules]
• All GAAP (Generally Accepted Accounting Principles) principles are rules that every enterprise must comply with. [Accounting Rules]

What follows is a cursory exploration of current business rules and future enterprise rules.

Business Rule — “As Is”

Part of the scope statement for the BRG (Business Rule Group) project — ”Defining Business Rules — What Are They Really?” — was stated as follows:

“The project has intentionally not dealt with entire categories of issues pertaining to the behavior of people in an organization. These include:

• Capturing the softer rules that involve the use of human judgment.
• Looking at the other components of systems that are related to soft rules, and seeing how they come together to support the business rationale.
• Showing workflows and processes in terms of their relationships to rules.
• The process of acquiring, maintaining and enforcing rules.
• Other systems of classification, such as mandates, policies, guidelines, corporate culture, industry rules, and corporate rules.
• Determining when rules are ineffective.”

Based on the above exclusion list, it seems that business rules as currently defined are of limited use in decision making.

Enterprise Rule — “Ought To Be”

Peter Drucker, management guru as well as father of Management Science, coined the term “knowledge worker” back in 1959. To survive and prosper, the enterprise relies on its knowledge workers for agility and innovation to meet with competitive challenges, changes in customer demands, and capabilities in mass customization of products and services. Knowledge can be categorized into explicit and tacit. Solving complex business decision problems or being innovative depends more on tacit knowledge than explicit knowledge. Tacit knowledge is carried in people’s minds — “walking encyclopedia”. There is a time limit on how long an enterprise can hang onto the tacit knowledge. It needs to internalize available tacit knowledge in a series of inter-connected domain specific knowledge bases. A knowledge base, which provides decision support capabilities, consists of an ontology, facts, and an inference engine. The ontology and inference engine of a robust knowledge base are constructed from a combination of the following classes of rules.

• Definitional — e.g. term and description of a concept in natural language.
• Syntactical — e.g. construction of a formal reference framework.
• Semantic — e.g. interpretation of derived result with a given syntax.
• Taxonomical — e.g. classification of animals into mammals and birds.
• Associative — e.g. combination of known concepts to form new ones.
• Deductive — e.g. application of predicate calculus to make inferences.
• Inductive — e.g. from specialization to generalization through analogy.
• Mapping — e.g. equivalence of International GAAP to US GAAP and vice versa.
• Transformational — e.g. shaping of raw materials into product parts.
• Decomposition — e.g. dividing a whole into parts.
• Synergetic — e.g. aggregation of parts to form a whole.
• Transitional — e.g. changing from one state to the next.
• Symbolical — e.g. chemical formulae, genomic maps, etc.
• Computational — e.g. optimization, game theory, econometrics, etc.
• Searching — e.g. branch and bound, pattern matching, heuristic, etc.
• Selection — e.g. what to collect and discard.
• Adaptive — e.g. neural networks and genetic algorithms.
• Fuzzy — e.g. dealing with the big grey area of the real world.
• Expert — e.g. conclusion drawn from partial information or facts.

It should be noted that the above list might not be exhaustive.

Assessment of Current State

The work done by the BRG and the OMG (e.g. SBVR) on rules is a good start, but it ought to be broadened in order to satisfy the higher level needs of an enterprise. Enhancements to existing rule engines and BPM systems will be required to provide more capabilities that facilitate the construction of different types of ontologies and inference engines.