In order to evaluate the concept mappings from TIPA to ArchiMate, we will perform a BWW (Wand and Weber 1993) analysis according to two criteria: completeness and clarity. The Bunge-Wand-Weber Model provides an ontological evaluation of grammars method, where we compare two sets of concepts to identify four ontological deficiencies:
1. Incompleteness: can each element from the first set be mapped on an element from the second? – The mapping is incomplete if it is not total.
2. Redundancy: are the first set elements mapped to more than a second set element? – The mapping is redundant if it is ambiguous.
3. Excess: is every first set element mapped on a second set one? – The mapping is excessive if there are first set elements without a relationship.
4. Overload: is every first set element mapped to exactly one second set element? – The mapping is overloaded if any second set element has more than one mapping to a first set one.
The amount of TIPA concepts that have no representation in ArchiMate defines the lack of completeness. Clarity is a combination of redundancy, overload and excess of concepts. Lack of completeness can be a serious issue while lack of clarity can make the mapping unidirectional and hard to reverse.
Considering all the above, we can say our mapping is complete, because every TIPA concept has an ArchiMate representation of itself. Furthermore, ArchiMate concepts can be so generic in a way that can accommodate some TIPA concepts, meaning sometimes the mapping does not reflect exactly the actual element meaning, but its generic meaning. We take advantage of this, through a set of assumptions, in order to achieve completeness. However, an extension to specialize and accurately represent these concepts would be much welcomed.
As for redundancy, there is sometimes more than one ArchiMate element to represent a TIPA concept. This happens in Generic Work Product and Specific Work Product, because TIPA describes these concepts as a set of artifacts that result from performing each process. These artifacts are the ITIL elements of each process which were already mapped by Vicente and each one represents a different ArchiMate element. The problem with redundancy is that the “correct” ArchiMate concept has to be chosen according to context and experience, and although this choice is rather easy for human architects, it can be a serious problem for automated model transformations. So we conclude that our mapping is redundant.
We also find excess, as ArchiMate has concepts that are not defined on TIPA as stakeholder or principle. One could argue that implicitly they actually exist with their ArchiMate definitions, where stakeholder is “the role of an individual, team, or organization (or classes thereof) that represents their interests in, or concerns relative to, the outcome of the architecture” and principle “a normative property of all systems in a given context, or the way in which they are realized” or even gap “an outcome of a...