I have read your comments and have no fundamental objections to your statements. I believe that we are in agreement that semantic precision is much desired, is achievable by sufficient rigour in modelling and is sometimes missing in STEP constructs. We seem to agree, too, based on your paper, that some concepts are needed as axioms, otherwise you just argue about the meaning of words and end up in infinite regress. My remarks were meant as a caution to avoid that.
Now the remaining issue is: Why and how are STEP definitions semantically imprecise? I would distinguish two main cases: Weak logic and imprecision by design intent.
Weak logic and poor wording does happen in STEP entity definitions. Sometimes constraints are very weak or missing so that the entity can mean much more than the verbal description says. This is a weakness in modelling style, where we agree that this should be corrected.
The other case relates to generic concepts at the resource level of STEP like 'product' or 'action', where you criticise the lack of semantic depth. I claim this lack is by design intent. If you look at STEP Parts 1 or 41 (fundamentals of product description), you will find a hierarchy of levels of generic like:
At the product_definition level you just assert that a product exists. You might also call it item, and its attribute is an ID.
At the product_property_definition level you assert that certain properties exist.
At the next level you define how the properties are to be represented.
Thus the semantic commitment increases, as Chris Partridge phrased it I believe, as you go from level to level. Finally you will specialise the generic concept in some application context, which is where you can now expect to see semantic precision.
This layered approach in STEP for achieving complete definition in multiple steps is by design intent. Thus your criticisms with respect to 'product' should be seen in this context.
ProKSI-97
Report