guest post by Jan Purchase, Director of Lux Magi Ltd ~~~~~~~~~~~~~~~~~~~~~ I do see value in distinguishing ‘policy’ rules from ‘expert’ rules. It’s also clear that, at their extremes, these rule types are poles apart. Still I fear that this distinction may be a continuum rather than a discrete dichotomy. Policy rules might be exemplified as the constraints of business operational practice – the rules that dictate what a company should and should not do. They might be mined from the boundary conditions of the fact model of the company’s business capability – as you suggest in your excellent book on this subject[1]. An example policy rule might assert, for example, “A company agent may be assigned to a high-value customer only if the agent is assigned to at most five other clients”. Expert rules, on the other hand, are often seen as more complex – for example, using heuristics to determine the best, anti-cancer drug to apply in a given situation, or forecasting regional sales opportunities. Often these expert rule bases use expert experience, genetic algorithms, fuzzy logic and feedback loops to reach a decision. They seek to augment, or even supplant, the wisdom of a small population of subject matter experts. But isn’t there a middle ground too? Consider an insurance policy decision rule that bin-sorts clients into good, bad and medium risk (the latter to be referred to a human underwriting expert). If this were a simple four-row, decision table based on the value of the policy and the risk profile of a client, you would probably consider it a ‘policy rule’. But if I add scorecarding, heuristics and analytics, at what point does it become an ‘expert rule’? Would you consider all decision rules (as opposed to constraint rules) to be ‘expert rules’? Or do they need to be complex or directly represent the experience of SMEs? Don’t all rules, to some extent, represent the wisdom of experts? Don’t many of them constitute and inform policy? In short: Is there a simple question we can ask (concerning a rule) to determine if it is a ‘policy’ or ‘expert’ rule? ~~~~~~~~~~~~~~~~~~~~~~~~ See my reply to Jan: http://www.brsolutions.com/2012/05/29/the-debate-continues-expert-systems-vs-business-rules/ Answer to Jan’s Question: The question to ask of every rule is whether the end-point is enforcement or is it a decision. An enforcement rule never becomes a decision rule, and a decision rule never becomes an enforcement rule. Once an enforcement rule, always an enforcement rule (assuming you don’t retire it). You can adjust thresholds (e.g., the mph of the speed limit), you can change the enforcement level (e.g., from ‘strictly enforced’ to ‘override with explanation’), you can change the sanctions (or eliminate them), etc., etc. And you can and should use analytics to measure and improve the rate of success in achieving underlying business goals. But it’s black and white. Enforcement is enforcement and decisions are decisions.

Here is my latest post in the on-going debate over decision management systems, expert systems, and business rules. ~~~~~~~~~~~~~ There is a fundamental difference between rules whose intent is enforcement (however strict) vs. rules whose intent is to make (expert) decisions. Rules whose intent is enforcement (e.g., speed limits) revolve around:

* detection of violations (think speed trap) * level of enforcement (e.g., strictly enforced) * violation message (electronic sign flashes ‘You’re speeding’) * violation response (cop chases you down the street with siren) * sanction or penalty (speeding ticket and a fine)

I chose an example that is probably not automatable (never be too sure) because such ‘behavioral rules’ (SBVR term) are everywhere in everyday life and therefore easy to comprehend independently of existing platforms and IT support. But there are a huge number that are automable; we just seem to be blinded to them sometimes for whatever reason (probably technical bias). Behavioral rules would not be involved in diagnosing (deciding) what’s wrong with a missle or classifying (deciding) the risk category of a prospect for insurance. In SBVR those are ‘definitional rules’ (or you could call them decision rules). They are about (encoding the know-how to make) smart (expert) decisions. It is true that decision rules often support behavioral rules in some fashion (e.g., is this particular speeder worth bothering over?). But it always comes down to this fundamental distinction: Is the end-point about enforcement, or is it about a decision. Enforcement and decisions are simply different. Are decision rules and behavioral rules both business rules? Yes. Should they be treated the same by platforms and methodologies? No. Why? They are different. Failing to understand the difference harms both business ‘users’ (poor governance processes) and decision management systems (oversell).

guest post by Ryan Trollip, Practice Director, Decision Management – www.prolifics.com ‘Expert system’ covers a pretty broad swath. Rules engines in the business world, in practicality, and in the majority of implementations, are simply operationalizing decisions, whether derived by predictive modeling or prescriptive business rules (e.g., regulatory). The conditions that reach a decision are largely pre-determined and operationalized in the rules system. Yes, there is a RETE algorithm involved. But don’t be fooled, this doesn’t give it intelligence, it is simply a style of execution. You can argue that in some implementations, sequential (non-forward chaining) is sufficient. In the real world, the management tools for rules systems, in my opinion, are more important than the algorithm. They have become the focus to externalize rules and allow for rapid change. I wouldn’t call a business rules system an expert system although you could probably create one with the tools out there. It’s simply a specialization much like how DBMS came about to better handle data. Not as sexy as AI, imperfect reasoning, etc., but certainly useful and practical.

guest post by Hafedh Mili, Professor of Computer Science at University of Quebec Montreal and Computer Software Consultant I agree that expert systems and business rules tackle different problems. Expert systems tackle two problems:

(1)   Difficult problems with no exact solutions – requiring heuristic knowledge.

(2)   Scarcity of expertise.

Hence, their first applications tackled complex engineering (and medical problems). Business rules tackle different challenges with respect to common business knowledge:

(1)   Externalization (explicitation).

(2)   Uniformization.

(Yes, I did just make up some new words.) There was no algorithmic/procedural INTERNIST that preceded the expert system INTERNIST. In contrast, plenty of business information systems implement business rules today – just not in the proper, agile way. So why did expert systems fail? Actually, did they fail? Was it a technical/scientific problem, or a business (model) problem? One could argue that expert/knowledge-based systems failed to solve the fairly challenging technological-economic problem of building true expert systems in a cost-effective way. What the science has been able to build is idiot-savants – i.e., systems that manipulated symbols and churned out ‘decisions’ without ‘understanding’ what they were manipulating. When INTERNIST is told about an old Chevy that overheats and has brown/reddish spots on its body, it comes up with MEASLES as a diagnosis. To make such systems less brittle, they need common sense. Doug Lenat (instigator of the CYC project) wittingly characterizes such common sense as “the things you need to know that enable you to disbelieve every word you read in ‘the News of the World’ (i.e., tabloids)”. Building such a common-sense knowledge base is not only costly, but epistemologically (and from an engineering point of view) a far more difficult problem than, say, codifying the debt-over-income requirements for applicants seeking multi-borrower mortgage loans. So, business rules vs. expert systems? Same mechanics (rules and rule engines), but significantly different problems. ~~~~~~~~~~~~~~~~~~ My reply: Hafedh, Great response. Not sure about the ‘same mechanics’ though. But that’s a post for another time.

The guest post by Cecilia Pearce earlier this week (http://goo.gl/QL9zL) stirred up an unexpected controversy, one that deserves clarification – are business rules and expert systems the same? No!  Business Rules  Business rules are organizational rules created for the purpose of running day-to-day business operations. Business rules always have an original source in the form of some law, act, statute, regulation, contract, agreement, business deal, business policy, license, certification, service level agreement, etc. I often like to say that business rules are really about keeping commitments. Expert Rules  In response to Cecilia’s post, Rolando Hernandez explains:

 “If you need to go beyond gathering [i.e., harvesting or mining] business rules to trying to understand “The Knowledge” [sic] that experts know, how experts think and decide, what the expert rules are, or what the higher-level heuristic rules are, then knowledge engineers … will keep using “knowledge acquisition” (KA), “knowledge representation” (KR), and “knowledge engineering” (KE). AI guys know that means.”

In other words, expert rules arise from an individual who is outstanding at his particular knowledge task. That’s very different. Expert Systems  Wikipedia describes expert systems as follows: 

 “software that uses a knowledge base of human expertise for problem solving, or to clarify uncertainties where normally one or more human experts would need to be consulted … a traditional application and/or subfield of artificial intelligence (AI)”

Bob Whyte, a practitioner for a major insurance company, makes the following observation about the difference between business rules and expert systems[1]:

“What makes the real-world challenge of managing business rules so much more tractable than it appeared to academics and researchers in the1980s, the heyday of knowledge engineering and expert systems, is that in the day-to-day business world the institution plays role of ‘god’. 

… for business rules the problem is not one of having to discover and define hidden, unknown or unexpressed rules, which takes you into byzantine solution spaces, but rather one of documenting known rules invented overtly and explicitly by actual historical person(s). 

With business rules you are generally not discovering rules no one has ever consciously considered, but rather uncovering rules that some manager, lawyer or other expert decided on one day, but probably did not record simply for lack of an appropriate infrastructure for rulebook management.”

Excellent clarification!

---

[1] from Building Business Solutions: Business Analysis with Business Rules, by Ronald G. Ross with Gladys S.W. Lam, An IIBA^® Sponsored Handbook, 2011, pp. 257-258http://www.brsolutions.com/bbs 

There are very important differences in the traditions of business rules vs. expert systems. Perhaps that’s why business rules had a completely different origin. In any case, they didn’t start finding each other until the late 1990s. (The first Business Rules Forum Conference was in 1997 … and every year since except in 2000.) The general goal of expert systems was broadly to mimic intelligent behavior – any kind of intelligent behavior. As a colleague put it, that’s like trying to read the mind of God. Human behavior (even the not-so-intelligent kind) is exceeding complex. The goal of business rules was always to capture the rules of organizations, not individuals. That’s one or two or more orders of magnitude easier – those rules have to come from somewhere … and that ‘somewhere’ was originally knowable (even if an arbitrary design decision by some programmer). So the issue with business rules is as much about business traceability (rule management) as it is expression. This problem goes right to the very heart of business governance and business agility. Continuing to embed business rules in procedural code (a) makes the business rules very difficult to trace, and (b) very difficult and expensive to change. It’s like setting the rules in concrete. It also precludes the possibility in the future of supporting specification-time detection of anomalies and intelligent dialogs to help Business Analysts remove ambiguity. For business rules, it ultimately comes down to the words you use and ‘remembering’ the interpretations made of them. There’s no way to demonstrate compliance without words, and no way to support transparency and accountability without traceability. The solution to all these problems, which are problems of business governance and therefore business engineering, leads inevitably in one direction. That’s why I’ve put so much time into researching business rules since the early 1990s and before. (Originally we thought in terms of databases and integrity constraints, another difference in origin from expert systems.) It’s also why I’ve spent so much time on the SBVR standard over the past 6 years or more. (At the risk of oversimplifying hugely, SBVR is about words and sentences.) The bottom-line: The way we do things today simply has to change. Change does take time. It also takes false starts and trial-and-error. But we’ll get there. Hey, business automation is barely one human generation old. That’s incredibly fast in the big scheme of things. ~~~~~~~~~~~~~~~  Read more about the history of business rules: http://www.brcommunity.com/history.php