Given the glut of redundant, overlapping, circular, or conflicting definitions, it may help to remember that “define” literally means putting limits upon. Definitions and their targets are two different things, the former being language constructs (intensions), the latter set of instances (extensions). As a Chinese patriarch once said, the finger is not to be confused with the moon.
In order to gauge and cut down the distance between words and world, definitions can be assessed and improved at functional and semantic levels.
What’s In & What’s Out
At the minimum a definition must support clear answers at whether any occurrence is to be included in or excluded from the defined set. Meeting that simple condition will stamp out self-sustained semantic wanderings.
Since definitions can be seen as a special case of non exhaustive classifications, they can be assessed through a straightforward two-steps routine:
- Effectiveness: applying candidate definition to targeted instances must provide clear and unambiguous answers (or mutually exclusive subsets).
- Usefulness: the resulting answers (or subsets) must directly support well-defined purposes.
Such routine meets Occam’s razor parsimony principle by producing outcomes that are consistent (“internal” truth, i.e no ambiguity), sufficient (they meet their purpose), and simple (mutually exclusive classifications are simpler than overlapping ones).
Functional assessment should also take feedback into account as instances can be refined and purposes reconsidered with the effect of improving initially disappointing outcomes. For instance, a good requirements taxonomy is supposed to be used to allocate responsibilities with regard to acceptance, and carrying out classification may be accompanied by a betterment of requirements capture.
Once functionally checked, candidate definitions can be assessed for semantics, and adjusted as to maximize the scope and consistency of their footprint. While different routines can be used, all rely on tweaking words with neighboring meanings.
Purposes & Capabilities
On a broader perspective, definitions can be ranked with regard to purposes and capabilities:
- Lexicon: flat and non specific list of words.
- Thesaurus: cross and domain specific semantics of words.
- Ontology: cross and domain specific semantics of concepts with epistemic qualification of whatever is considered.
- Models: cross and domain specific semantics of concepts with epistemic qualification of whatever is considered and rules to be applied to the processing of representations (descriptive, predictive, or prescriptive).
That principled approach can be used to clarify the scope and reliability of competing standards. It could also be extended to the design of ontologies.
Ontologies & Business
Ontologies are all too often seen as abstract contraptions best reserved for arcane issues. But, as noted above, ontologies are meant to be built on purpose, to flesh out thesauruses with actual contexts and concerns, and put to use.
That could help enterprises confronted to the crumbling of traditional fences, changing business environments, and waves of digitized flows with confusing semantics.
To manage these challenges enterprise governance need knowledge architectures bringing together heterogeneous and changing business contexts as well as homogeneous and stable models of organization, systems, and platforms; that cannot be achieved without open and modular ontologies.
But for that to be achieved, means, e.g conceptual graphs or semantic networks, should not be confused with ends, i.e the purpose of ontologies. Whereas implementation issues are not to be ignored, the priority should be to characterize ontologies with regard to the social basis of contexts (institutional, social, professional, corporate, personal), and the epistemic nature of targeted instances (concepts, documents, actual occurrences, or symbolic representations.)