Excerpt from Enterprise Architecture Fundamentals:
As far as systems architecture is concerned, modeling languages provide principled and effective schemes that bridge the divide between business analysts’ requirements and software architects’ designs. But enterprises’ endeavors and behaviors cannot be fully fitted into systems-oriented modeling categories, and pressing on in that direction would hamstring the ability of enterprises to anticipate changes in business environments. The immersion of enterprises in digital environments reinforces the inhibiting impact of looking at business environments through systems-modeling glasses:
- Facts, once captured through discrete observations and ready to be interpreted, have turned into raw, massive, and continuous data floods swamping predefined categories.
- Concepts, once built on explicit models and logic, are now emerging like new species from the primordial soup of digital environments.
Typically, business analysts take the lead extending symbolic representations on both fronts: toting learning machines for facts and waving knowledge graphs for concepts. Positioned in the no-man’s land between fast-flowing facts and shifting concepts, systems architects have to deal with a two-pronged encroachment on their information models:
- Regarding facts, systems architects have to build a Chinese wall between observed data and managed information in order to comply with privacy regulations.
- Regarding concepts, they have to continuously integrate emerging concepts with the categories managed by information systems.
That brings a new light on the so-called conceptual, logical, and physical “data” models, which are key components of enterprise architecture:
- Physical data models are meant to be directly lined up with operations and digital environments.
- Logical models represent the categories managed by information systems and must be congruent with their functional counterpart in systems architecture.
- Conceptual models represent the enterprise’s explicit knowledge of business domains and objectives; such models should also be amenable to the implicit knowledge embodied in people and organization or elicited from environments through Deep learning.
Logical models (pertaining to information) serve as a modeling hub between business facts (observed through data) and concepts (used as knowledge axes or pillars), securing the adjustments of representations and environments; typically (figure 4-10):
- Knowledge is used to analyze data sets that could identify and inform prospective business categories (a).
- Knowledge uses business categories (information) to build new marketing data sets (b).
- Data sets are crossed with business categories (as managed through information models) in order to assess and improve marketing knowledge (c).
With everything turning digital, these distinctions could be easily overlooked were it not for the extensive adoption of Knowledge graphs and Deep-learning technologies. Digital strategies cannot be built without a clear understanding of digital contents; that’s a prerequisite.
Then, in order to compensate for digitally levelled playgrounds, enterprises need a discriminating and transparent governance apparatus.