Comments to the UCR FPWD

Dear all,


Let me congratulate and thank you all for your work. The mailing list is very active and the results can be well seen in this FPWD. Although I subscribe the mailing list, I have not been following all the discussions, so I apologize if any of my thoughts duplicates any of your previous discussions. Please use what you think may be useful in the context of your work. And, of course, I am available to further discuss these issues.

Thank you and best regards to all.

Ana 


--------------------------
In the Introduction:


“In addition to the variety of standard vocabularies, there have been multiple definitions of application profiles, which define how a vocabulary is used, for example by providing cardinality constraints and/or enumerated lists of allowed values such that data can be validated. “

I do not agree that application profiles “define how a vocabulary is used”.

I think we have here a good opportunity to clarify concepts and to eliminate some misunderstandings and misuses of the term Application Profile. In my view, the concept is larger than the term that we are currently using. I would prefer to use Metadata Profile instead of Application Profile, because it better expresses the meaning the community is giving to this concept. Under this view, I can think of, at least, three kinds of metadata profiles:

-       Application Profiles – designed to meet the needs of a specific application, as argued by Heery and Patel [1].

-       Community Profiles – designed to meet the needs of specific communities, but with no specific applications in mind. These will probably need to be reshaped when instantiated in specific applications.

-       Dataset Profiles – designed to fit the needs of one or more datasets. This seems to me the trickiest one as here I am referring to datasets that already exist and some agent (person, organization,...) wants to make open as Linked Open Data.


Vocabulary Profiles would be a reshape of a single existing vocabulary designed to meet the needs of a specific application, community or dataset. A Vocabulary Profile would be orthogonal to the other three kinds of metadata profiles. 


I would say a Metadata Profile expresses how one or more terms from one or more vocabularies are used and related in a specific context. That context may be a specific application, in which we call the profile a Metadata Application Profile.

 

[1] http://www.ariadne.ac.uk/issue25/app-profiles

 
-------

I don’t know if I still can propose Requirements. If yes, please consider the following:


5.51 – Multidimensional Metadata Profiles

Tags: Profiles

Stakeholders: data consumer

Description:

Similar data generated by different providers will likely arrive in different formats and with different properties and restrictions. When modelling particular applications that consume such diverse data, one should consider all these different settings and specify what to do with each of them. For each attribute defined in the domain model, it should be possible to specify all the related metadata properties used by the different datasets, their restrictions and relations. There should also be a canonical view to which all equivalent properties would map.
 

5.52 – Arrangements of controlled vocabularies as ranges

Tags: Profiles

Stakeholders: data producer

Description:

In a Metadata Profile it should be possible to define as range of a property an arrangement of one or more controlled vocabularies (subset, union, intersection,..).

---

Ana Alice Baptista
______




ALGORITMI Research Centre / Information Systems Department
School of Engineering, University of Minho
P-4800-058-Guimarães