Semantic Modeling for Data

Chapter 14. Evolution and Governance Dilemmas

It is not the strongest or the most intelligent who will survive but those who can best manage change.

Leon C. Megginson

Semantic models are dynamic artifacts that need to evolve over time if they are to maintain and improve their quality and usefulness. Typical model evolution tasks include fixing quality issues, adding structure and content to cover additional domains or applications, or removing elements that are no longer valid due to semantic change.

Moreover, as we saw in Chapter 11, it’s very important that this evolution is done in a controlled and strategy-compatible way, meaning that an effective model governance system needs to be in place. In this chapter I describe some key dilemmas related to these two major tasks and discuss ways to tackle them.

Model Evolution

Evolution is a very important aspect in the life cycle of a semantic model that you can only avoid caring about if:

The model’s first version has the quality you need/want in all the relevant dimensions

If this first version has quality problems in one or more relevant dimensions (accuracy, completeness, etc.), you will need to fix these problems in subsequent versions.

There are no external forces that may cause this quality to deteriorate

Such forces include changes in the model’s existing domains (e.g., new knowledge being added or existing knowledge becoming invalid), scope (e.g., new domains or applications to support), and/or quality requirements ...