Conceptual Data Models versus Business Semantics

Business semantics seems to be beginning to gain traction in the areas that are often first adopters - defense and financial services.  The definition of "business semantics" is still a bit fuzzy, but it is generally recognized as centering on understanding the reality of business information without any concern for what is stored in databases.  The value proposition for business semantics seems to be coming from a number of different directions, including:

• A general lack of understanding of business information in many areas of the business
• Difficulty in communicating business terms across the enterprise
• Deformation of business information as it is put into databases
• Lack of management of information in databases after production implementation
• Loss of understanding of information in databases after production implementation

We can expect this list to get longer and clearer in the coming years.

Yet we have had conceptual data models for many years, and are these not the same as business semantics?

I think there really are some differences.  The first is that conceptual data models are essentially about important business entities, and contain only a few attributes.  By contrast, business semantics concentrates on pure concepts that can be entities or attributes, but are mainly the latter.

A second difference is that conceptual data models can have relationships and these can even be many-to-many.  In business semantics, however, relationships are much more precisely defined.   Furthermore, business semantics can have a wide range of different kinds of relationships, e.g. attribute-to-attribute.  From what I have seen, these varied kinds of relationship cannot be captured in a traditional conceptual data model.

Next, business semantics captures taxonomies, whereas conceptual data models do not.  A business concept such as Customer Credit Status can be defined in both conceptual data models and business semantics.  But the subconcepts of Customer Credit Status, like Platinum, Gold, Silver, and Bronze, cannot be captured in a conceptual data model.  Yet they can, and must, be captured in business semantics.  The ability to define and manage taxonomies, in fact, is something that does not happen in logical or physical data models either.  It is a crucial advantage of business semantics.

Then there are rules.  An arbitrary number of rules can be captured in business semantics.  These can also be documented in conceptual data models.  However, they are typically regarded as too detailed for conceptual data models and are more often found in logical data models.  From what I have seen of business semantics, the treatment of rules may still be somewhat immature, so the differences with conceptual data models may not be material.

This is just an overview of the treatment of metadata within the two classes of model.  There are also wide differences in governance and methodology.  Given this, it is legitimate to ask if conceptual data models and business semantics serve different purposes.   Sometimes conceptual data models are a starting point for developing a physically implemented database, but sometimes they are claimed to show the reality of business information.   Given the advances in business semantics, I think it is going to be increasingly difficult to make the second of these claims.