I was looking through my notes about articles that I had read - and found an interesting Burton Group report entitled Generalized and Detailed Data Models: Seeking the Best of Both Worlds. (I think that it was published earlier this year.) I must admit to having been both confused and intrigued by the title. :-)
In the paper, "generalized" models are those used to define database/storage structures and to find the general themes and fundamental aspects of the data (and its values). In short, they are the data models defined by IT to effectively and efficiently use the technologies that are in place (like SQL databases). Maybe "reduced" is a better word than "generalized" ...
On the other hand, "detailed" models are those that are useful to business people. They define and describe the information requirements of the business, and its vocabularies, rules and processes. They hold the details from the business perspective. Again, maybe another word like "conceptual" is better (since even the "generalized" models hold "details") ...
What is valuable is not the titles used for these models but their semantics. :-) The key message is that a business needs both types of models and they need to stay in sync. This is really important. The conceptual/detailed models hold the real business requirements and language. They haven't been reduced to basic data values whose semantics are lost in the technology used to define and declare them.
IMHO, a business loses information and knowledge when it only retains and works from the IT models. There is much to be gleaned from the business input and much value in keeping the business people engaged in the work. This is almost impossible once you reduce the business requirements to technology-speak.
As the report says, "do not allow generalized models to compromise your understanding of the business."
Posts
Coronavirus and Machine Learning Conferences
4 years ago
I think some of the modeling has characteristics that transcend the application domain… A concept that you get, but several miss.
ReplyDeleteExample, ISO 15926 "Industrial automation systems and integration—Integration of life-cycle data for process plants including oil and gas production facilities" is a standard for data integration, sharing, exchange, and hand-over of industrial processing plant information between computer systems.
DMTF’s CIM and WEBM are dealing with standards for systems management in enterprise IT environments.
They’re both LIFECYCLE INVENTORY SYSTEMS and have a lot in common, but with domain-specific implementations. It would appear that SOME shared use of common business vocabulary and tooling would make sense. Even better, semantic web concepts and technology should be build into standards implementations. ISO 15926 is well on its way on that count.
Yes, confusing terminology in that report! And some good points.
ReplyDeleteA domain or conceptual model is so valuable -- it formalizes what it is you're talking about and building, and it can expose inconsistencies in your semantics before it's expensive or klugey to work around. I truly believe that modeling your domain before you model your data results in better designed, more extensible and maintainable systems.