Continue in 2 seconds

Hitchhiking to a Semantic Model

  • April 27 2006, 1:00am EDT

This month's column is contributed by John Schley, president, DAMA International.

A few years ago, I had the good fortune to develop an enterprise-wide semantic model - except I didn't know that's what we were doing.

Coming from a data management background, I called it a conceptual data model. A large project to reinvent our mortgage company's loan servicing system was kicking off, and I was new to the business. As a way to compensate for my lack of business knowledge, I asked the project team to develop a conceptual data model of the major entities in that space. This turned out to be more complicated than it may seem, because the different parts of the company used the same terms differently. Was it a loan when it was applied for ("application"), signed ("closed") or when we had completed our final review of all the loan documentation ("post-closing")? The answers varied, depending upon who was doing the answering. As we developed our conceptual data model, we came together to concisely define each entity. We kept the process moving, and in a few weeks, we had identified and defined approximately 50 business concepts in our "concept model." I use the term concept model, because term "conceptual data model" didn't make sense to the business partners. They didn't see this as focusing on data but on their business terms. Using their term increased their sense of ownership in the effort.

A year or two after that, the organization adopted an object oriented (OO) approach to systems development. The OO developers wanted a high-level class diagram that showed the major objects in the enterprise so they could begin building components around them. Rather than start a new effort to search these objects out, they recast the concept model developed earlier as a class diagram and drove discovery and documentation of another hundred or so classes/entities.

Not long after that, the company was sold to another mortgage company. In preparation for the conversion to the new company's systems, a large-scale effort was made to document all our data structures. The data management staff and I used this opportunity to not only document our physical data models but leverage it into a way to create a home-grown enterprise logical data model by "logicalizing" the physical data elements into a single logical view. This "bottom-up" effort resulted in a logical model that encompassed approximately 75 percent of the company's structured data.

Over this time, we had developed a set of semantic models that integrated data across the enterprise. Along the way, we had filled in the cells in the data column of John Zachman's Framework for Enterprise Architecture and built links between the different levels to associate different perspectives of the same data. It had morphed from a concept model to a class diagram to an attributed logical data model, hitchhiking on different projects with different goals. The final product was driven by different efforts but remained intact to aid the organization's future development.

High-level semantic work is valued by the business, although it must be given a name that is meaningful to the business and it must meet a business need. This kind of work cannot be prescribed by information technology, project management or by anyone's "best practices," but must relate directly to an existing business problem and be substantial enough to help solve that problem. When those pieces are aligned, it's time to stick out your thumb and hitch a ride to a semantic model.

Check out a DAMA International chapter meeting near you to hear experiences like this and learn about data management from other practitioners and thought-leaders in the field. Visit our Web site at to learn more on how to get involved with DAMA.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access