Re: comments on section 7.4

Hi Joao Paulo,

Thanks a lot for the feedback!


>
> The problem is that in the example the data itself does not become "richer". There is a richer description of data (localized description)... let's replace by something more neutral, say,
>
> instead of "They can also provide a way to easily create richer data."
>
>   we could say "They can also serve to facilitate the use of data."  or "They can also serve to improve the usability of datasets"... than the example will make more sense and we will not be calling data itself "richer" without explaining what that would entail.
>

I'm not sure what the nuance is with "description of data". I live in a sector in which metadata (say, about books) is data, and it becomes richer when it's linked to vocabularies.
Anyway, your suggested sentence is good, and still conveys the point, so I've just put it in!



>
> Great. There is just a typo in the text that you added for this ("vidual"->"visual"). We can discuss later whether to add the diagramming conventions.


Typo fixed!

  
[About the BP on formalization levels]
>
> Thanks Antoine, I think you did a great "weaving" of our perspectives in the text.
> "Intended outcome" and "How to test" are still biased. Perhaps we can leave it as is for now and add an issue to investigate this in the future.
>
> I am finding it difficult to understand why the relevant requirements for this BP are: Relevant requirements: R-VocabReference <http://www.w3.org/TR/dwbp-ucr/#R-VocabReference>,


If a vocabulary is too complex for vocabulary re-users, then it will be hard to position it as a reference vocabulary. It will have to 'compensate' by showing a very appropriate fit to the applications at hand.


> R-VocabDocum <http://www.w3.org/TR/dwbp-ucr/#R-VocabDocum>,


Good formalization helps documentation (because it makes meaning more precise) as you pointed out.
Complexity of specification on the other hand has a direct impact on the ease of producing and understanding documentation.

Compare the doc for a relatively simple vocabulary:
http://www.w3.org/TR/vocab-org/
(this includes all classes and properties, and all formal axioms that apply to them)
and what happens a much more complex one:
http://www.w3.org/TR/prov-constraints/
(this doc contains only for the formal constraints)



> R-QualityComparable <http://www.w3.org/TR/dwbp-ucr/#R-QualityComparable>
>


If two vocabularies have very loosely defined meaning, ambiguity may make it more difficult to align them and thus to compare the data expressed with them.

If two vocabularies are extremely formalized it may result in the same notions being given different (very precise) semantics. If this specific formalization was not so crucial to their respective applications, then we've introduced unnecessary obstacles to the comparison of the data expressed by the two vocabularies.

Note that all this relates to following the way to what Tom Gruber called "Minimal ontological commitment" in his ontology papers [1].
The original BP title I had chosen (on "overformalization") was a reference to Gruber's notion of "overcommitment".

Best,

Antoine

[1] T. R. Gruber. Toward principles for the design of ontologies used for knowledge sharing. 1993
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.91.6025

[1]

Received on Friday, 23 January 2015 13:30:54 UTC