Rules, Business Rules, and Big Data: What’s It All About?It’s time to come to grips with what is meant by “rule” in the context of big data. There’s much confusion out there. In a recent keynote, Rick van der Lans stated, “… big data leads to more and more interesting insights, and from there to more and more rules.” What does he mean? The funny thing is you can also call ‘insights’ rules … and some people do(!). Not me! Read on. An Example One of Rick’s examples of rules born from big data:
If 2 calls disconnect within 10 minutes, then offer a product discount.What’s the insight and what’s the rule? Does the statement represent both? Does it express a business rule? The syntax of the statement is in if-then form. Doesn’t that imply a business rule?! No! According to the standards SBVR and the Business Motivation Model (BMM), business rules must be:
- Declarative. The statement above is not declarative because it includes the command “offer”.
- Practicable. The statement above is not practicable – not ready to roll out into prime-time business operations – because it’s ambiguous. More on that momentarily.
- “2 calls disconnect within 10 minutes” … That part of the statement suggests an insight: Calls on hold for 10 minutes or more are likely to disconnect.
- “offer a product discount” … That part of the statement suggests a remedy, a way to recover from a bad situation.
We can assume people are getting frustrated at the 10 minute mark or before. If we offer a product discount, they’ll be mollified and more likely to hold on or to purchase.What should we call the statement? It does give guidance and it does clearly have a role in strategy. However, neither the insight nor the remedy is practicable. Here are some unanswered questions that could produce ambiguity.
The insight part: Does the 10 minutes refer the wait period on each individual call? Or to any time interval during which calls are waiting?
The remedy part: How much discount? On which product(s)?So according to the standards the statement represents a business policy, not a business rule. A corresponding business rule might be:
A caller must be offered a 15% discount off list price on any product in stock if the caller has been on hold for more than 10 minutes.This version removes the ambiguities. It clarifies that we’re referring to:
- The wait period on an individual call.
- A 15% discount off list price.
- Any product in stock.
2a(1): a statement of a fact or relationship generally found to hold good : a usually valid generalizationLet’s call this meaning rule1. It roughly corresponds to insight … i.e., “as a rule we find that …”. It’s experiential – based on evidence.
1f: one of a set of usually official regulations by which an activity is governedLet’s call this meaning rule2. It roughly corresponds to the underlying sense of business rule … i.e., “It’s necessary or obligated that you must …”. It’s deliberate, based on policy. Job one in analysis of big data is to identify interesting relationships (rule1) and then deliberately formulate business rules (rule2) to produce outcomes desirable for your company. In other words, starting from rule1 you want to move expeditiously to rule2. Logicians have been on top of this distinction for a long, long. Only they speak in terms of implications, not rules. There are two kinds of implications – material and logical. Let’s repeat the discussion above using these terms. Don’t overlook the word strictly in the second definition. material implication (rule1)
2b(1) : a logical relationship of the form symbolically rendered *if p then q* in which p and q are propositions and in which p is false or q is true or bothlogical implication (rule2)
2b(2) : a logical relationship of the form symbolically rendered *if p then strictly q* in which q is deducible from pLet me repeat myself on job one in analysis of big data using implication:
Job one in analysis of big data is to identify material implications (rule1) and then deliberately formulate logical implications (rule2) to produce outcomes desirable for the company. In other words, starting from material implications (rule1) you want to move expeditiously to logical implications (rule2).I use “business rule” only for a statement of the rule2 variety, and only if that statement is both declarative and practicable. A statement has to prove itself to be a business rule – it’s only a pretender if it fails to meet the standards.
Tags: BI, big data, BMM, Business Motivation Model, business policy, business rule, business rule vs. business policy, declarative, implication, insight, practicable, remedy, rule, Rules in BI, rules in business intelligence, SBVR