The Pragmatics of Creating Concept Models
Extracted from Business Knowledge Blueprints: Enabling Your Data to Speak the Language of the Business, by Ronald G. Ross, 2020.
In creating a concept model, you can start in any of several ways, whichever best fits your circumstances. For example, you can start by identifying terms or concepts that seem central to miscommunication and so most badly in need of clear definitions.
Or you can create a list of terms most basic or most frequently used in business communications within scope. Automated tools for extracting terms from existing, well-written documents are publicly available, many free.
It's always an excellent idea to identify existing definitions from reputable business glossaries to prime and accelerate discussion. Such glossaries may be external or internal to the organization. Our experience is that relevant business glossaries exist far more often than you might imagine. They can give you an important head start.
Another approach to developing a concept model, better in many respects, is in conjunction with the actual authoring of textual business statements themselves (e.g., policy statements, product descriptions, regulation, etc.). We have done this for many years, for example, in conjunction with writing business rules. There's nothing like actual written business communication to sharpen awareness of disconnects among different business participants than having to write things clearly.
No matter what approach you use, always remember you're not actually starting from scratch. People already have concepts in their minds. The challenge is that those concepts are often not very precise and not necessarily shared.
Your Business Glossary
In concept modeling you will be defining your business vocabulary. It's fundamental to your business knowledge blueprint.
In organizing the entries in your Glossary, you may include not only definitions, but examples, notes, and basis (any source used for a definition). You may also want to record such things as descriptions, reference sources, graphic illustrations, etc. Glossaries also typically identify synonyms and preferred terms, if any, and may include equivalent terminology in one or more other natural languages.
In organizing your own glossary, include as much or as little as you need (but never less than terms and definitions!).
Examples, Examples, Examples
A factor critical for creating a concept model is having plentiful examples at your fingertips. Always know whom to ask for such examples and interact with those people frequently.
We once created an extensive concept model with a client for medical insurance. The question arose whether disease and medical condition were the same concepts. We're not doctors! Only by accessing lists of official examples could we determine they are not the same concept (at least for that business).
We strongly recommend identifying typical examples for each concept (as well as counterexamples) and then recording them for future reference. You'll be pleasantly surprised how helpful they can be in jogging memory and in confirming and explaining concepts as time goes by. Make recording examples a standard part of your approach.
How big will your concept model be? How many concepts? The answer is naturally it depends. We have created useful concept models of only several dozen concepts. And also ones with many, many hundreds of concepts. Practical size depends on:
- Defined scope. (Always define scope!)
- Complexity of the business knowledge.
- Degree of disambiguation capability desired.
What are your objectives? How pervasive and costly is potential miscommunication? What size body of text do you need to disambiguate?
To give you a rough idea, the largest concept model we have helped create addressed over 10,000 business rules in property and casualty insurance. It included approximately 40 neighborhoods (diagramming divisions of convenience), 500 terms, and 750 verb concepts.
The best advice: Start small, then see where you need to go from there.
Concept models scale in size quickly. Flip charts will do fine to get started. Beyond a certain point, however, you will want tooling to help you.
Naturally the tooling should be diagram-friendly. You don't want to constantly wrestle with diagramming features and functions, so the diagramming capability needs to be as robust and as user-friendly as possible.
Diagramming alone, however, isn't nearly enough. Creating a concept model is text-intensive work, especially but not exclusively for definitions.
To adequately support the work, you need a tool capable of building a business-friendly glossary that can be used directly on its own by business people and business analysts. The tool should provide complete synchronization between the text included in diagrams, and the text represented in the glossary. Comprehensive tracking and notation for terms within all text is also essential.
The full vision for business vocabulary is complete integration with document-authoring tools used by business people. Why shouldn't definitional criteria be embedded behind every business communication written, whether contracts, product/service specifications, governance or regulatory-compliance documents — or even requirements for automated business solutions? In the Knowledge Age, definitions should be as immediate as spellcheck.
An important caution: Most repositories currently available were built for software engineering and software developers, not for capturing business knowledge blueprints. They are generally neither as text-friendly or as business-friendly as they need to be for concept models. You are not engineering software. There is no point in force-fitting an ill-suited tool meant for IT. Don't fall into that trap.
# # #