The Business Rules Group's current work-in-progress is "Organizing Business Concepts," a conceptual framework (logical model, if you prefer) to help enterprises organize their business knowledge and develop a shared vocabulary. Like other BRG developments, one way to visualize it is that the model could be used to specify a database in which people could define the terms they use in their business, and store and cross-reference facts and rules based on those terms.

Shared vocabulary is important. Does 'Customer' mean the person who orders your product? The person the product is delivered to? The person who will use it? The person who pays for it? The person who put your product on the "approved list" so that it could be ordered? The departments they belong to in the company they work for, rather than the people as individuals? The company itself? Several of the above? All of the above?

Does someone become a customer simply by buying something from you? Or is 'customer' a status they have to attain, via references and credit checks and so on, before they are allowed to order anything? The answers to all these questions will vary, depending on your organization, and maybe even between departments in your organization.

Shared vocabulary becomes even more important as the context of business operations expands -- as enterprises come together by merger and acquisition, as they get into tightly coupled partnerships, as they outsource business functions to suppliers (who may deal with many different client organizations, with broadly similar but by no means identical vocabularies), as they provide disintermediated delivery channels, so that their customers interact with a Web or WAP service rather than a person.

In the Group's last workshop the focus was on 'fact.' Our previous workshop focused on 'term,' and the next one will focus on 'rule.' This progression supports what Ron Ross calls the Business Rules mantra, "Facts are built on terms, rules are built on facts."

The mantra is something of a simplification. Facts are built on concepts. We name concepts by assigning terms to them. 'Customer,' 'purchaser,' and 'buyer' may all be acceptable terms for the concept 'a person or organization who buys our products' (that is, they may be acceptable terms within one organization -- they might have different meanings in others).

Also, people don't deal with facts in the abstract. They deal with statements of fact. The same fact may be stated in different ways, for example 'customers buy products,' 'products are purchased by buyers.' Different statements may use different, but equivalent, terms. The word or phrase sequence may be different. They may be in different natural languages, 'Les clients achètent les produits,' 'Die Kunden kaufen Produkte.'

This issue -- what things are, as opposed to what they are called and how they are described -- has exercised us from the start of "Organizing Business Concepts." In my last column I harked back to student days and quoted Wittgenstein. This time I was reminded much more of Alice. (For younger readers, Alice, of Wonderland and Looking Glass fame, was a favourite among programmers in the gentler days before Star Trek).

A little before Alice met Humpty Dumpty, as quoted at the top of the column, she met the White Knight, who offered to sing her his song.

We, both in the Business Rules Group and out in the real world of business, often get into the same kinds of confusion.

The mantra "Facts are built on terms, rules are built on facts" should perhaps be "Facts are built on concepts, rules are built on facts. Concepts are stated in terms. Facts are stated in fact statements and there may be many fact statements for the same fact. Fact statements use the terms assigned to the concepts that participate in the facts that the fact statements represent. Rules are stated in rule statements and there may be many rule statements for the same rule. Rule statements use the terms assigned to the concepts that participate in the facts that participate in the rules that the rules statements represent, but not necessarily the same terms used in the fact statements for those facts. Oh, and it probably makes sense, for any given rule statement, that the fact statements for the facts that participate in the rule, and the terms used for concepts that participate in the facts, are in the same natural language as the rule statement"

This is not quite so catchy as the original, but it's an important part of what we are working on in "Organizing Business Concepts."

One problem is embodied in the example used earlier: "customers buy products." Is it a term definition? Does it say that "customer" is a term we use for anyone who buys our products? Is it a fact statement? Does it say that there is a specific connection between the concept "customer" and the concept "product" that can be named as "buy"? (We need to name it so we can distinguish it from other connections, such as "complain about" and "recommend") Or is it a constraint -- a kind of rule -- that says "only people who meet the criteria for being customers are allowed to buy our products"?

This is a problem of categorization. One of the group's 'set books' is "Women, Fire, and Dangerous Things" by George Lakoff, the distinguished cognitive scientist. It's about categorization.

The title is taken from the work of RMW Dixon with Dyirbal, a language spoken by the Ngadjonji people of Queensland, Australia, which classifies all objects in its universe into four categories. To speak Dyirbal correctly, you must precede each noun with its classifier. One of these is 'balan,' which includes women, fire, and dangerous things. It also includes dogs, the sun and stars, most snakes, some trees, shields and some spears, anything to do with water, and a whole lot of other things.

Dixon's work (as reported by Lakoff -- I haven't read the original) is both fascinating and frustrating. At first glance the Dyirbal classification system seems bizarre, but Dixon was able to provide some principles that make sense of it (I haven't got the space here, you'll need to read the book). But the "rules" that a Ngadjonji person would assimilate in growing up as a Dyirbal speaker, the rules for actually classifying things in the world, are different.

So, if you were told the Dyirbal category of something, Dixon's principles would explain to you why it was in that category. But if you didn't know what something's Dyirbal category was, Dixon's principles would not give you rules to decide it.

The Business Rules Group can't just stop at explanatory principles. We need to provide guidance on how to categorize the kinds of thing you will encounter in building the vocabulary for your business. We actually need to go further. We can't define a single, universal classification system, so our model has to support the specific rules you'll need for your business environment. And, if we do our job well, there should be guidance on how you can develop your rules.

We didn't finish the discussion during the workshop. It continued for some time after via the BRG internal web site. It got a little tense at times, and some of the problems were about vocabulary. We couldn't discuss things without using terms to refer to them, and some people had strong feelings about the use of some terms. Members of the group even went so far as to "politely disagree," which, in email discussions, is pretty strong language. Mercifully, nobody escalated to "IMHO" or "with respect."

We were victims of the very problem we were trying to solve. We couldn't agree on the vocabulary to use in the model we are developing to help people to define their vocabularies.

One example was (here's the problem -- what do I call it in order to describe it? The link?) the link between two concepts when they are joined together to make a fact.

I have to confess to getting this ball rolling. During the workshop I objected to using "participation" as our term for this concept. In a masterly display of podiatric autofusillation I had forgotten that we had used "participation" for this concept in our foundation paper "Business Rules ~ What Are They Really?" and I was one of the contributors.

However, once I'd started the discussion, the terms came thick and fast. Association, relationship? Too much IS baggage from object modeling and data modeling. We want people to think about business concepts. Collaboration? Has the connotation of intent. Connection, link? Too general, too vague. The discussion continued via email. One member suggested "commitment." Another strongly preferred "coupling" to "commitment" (which also took me back to student days. Back then, so did I).

We still haven't resolved it. A strange thing happened. After a lot of emails -- our busiest time saw 45 in three days, about a third of them about participation (or association or coupling or connection or whatever. You see how difficult this is when you don't have consensus on a shared vocabulary?) -- the discussion just petered out.

Which reminded me again of Alice. After a long conversation about words and meanings, Humpty Dumpty recited his poem, which went on until:

We'll have to revisit the open issues at our next workshop. Watch this space.

References

Alice through the Looking Glass -- many editions. A personal favourite is The Annotated Alice, Martin Gardner, 1999 Norton, ISBN 0-393-04847-0.

Women, Fire, and Dangerous Things: What Categories Reveal about the Mind, George Lakoff, 1987, University of Chicago Press, ISBN 0-226-46804-6.

Dyirbal: spoken by the Ngadjonji people of Queensland, Australia (includes references to Dixon's work) http://www.koori.usyd.edu.au/ngadjonji/

# # #