W3C home > Mailing lists > Public > public-lld@w3.org > March 2011

Re: Question about MARCXML to Models transformation

From: <PATRICK.LE-BOEUF@bnf.fr>
Date: Wed, 9 Mar 2011 16:10:57 +0100
Cc: patricia.riva@banq.qc.ca,martin@ics.forth.gr,maja.zumer@nuk.uni-lj.si,smiragli@uwm.edu
Message-ID: <OF1B302025.4C46823C-ONC125784E.0051E25F-C125784E.00536677@LocalDomain>
To: public-lld@w3.org
>The FRBR and RDA vocabularies can be defined in an
>ontologically tolerant manner, such that data which uses the
>models imperfectly -- or data about things to which the models
>imperfectly apply -- will not raise fatal exceptions when
>linked with data that may be simpler, vaguer, or simply based
>on different models.  Apparent misalignments, or contradictions
>to the logic of the models, or gaps in descriptions, should
>be flagged with nothing stronger than helpful error messages.

I think FRBRoo can prove quite useful here, as it is plugged to CIDOC CRM 
and CIDOC CRM is an ontology that accommodates both extremely refined 
descriptions and very rough ones. The "level 0" implementation of CIDOC 
CRM would read:
E1 CRM Entity _P1 is identified by (identifies)_ E41 Appellation
E1 CRM Entity _P2 has type (is type of)_ E55 Type
E1 CRM Entity _P3 has note_ E62 String
which means: in my database, there is some stored information (E62 String) 
relating to some stuff (E1 CRM Entity) associated with some identifier 
(E41 Appellation) and that I can characterize as belonging to some very 
broad category (E55 Type, e.g. "library stuff").
Of course this is not the cleverest way to implement CIDOC CRM, but in 
some desperate cases it can be envisioned...
All FRBRoo classes are ultimately declared as subclasses of E1 CRM Entity, 
so this very vague set of assertions applies also to anything that belongs 
to the scope of FRBRoo. Of course, E71 Man-Made Thing, which I mentioned 
in my previous message, also is a subclass of E1 CRM entity, so it is 
possible to specialize these three assertions to just E71 Man-Made Thing 
as an approximation for the "WEMI whatchacallit."
Of course too, there is a very wide range of possibilities for FRBRoo and 
CIDOC CRM implementation, from that caricatural non-implementation to a 
pointillistic, extremely detailed use of all classes declared in FRBRoo 
and CIDOC CRM and a maximal use of all their expressive strengths.
Patrick









Message de : Thomas Baker <tbaker@tbaker.de> 
                      09/03/2011 04:00

Envoyé par : 
public-lld-request@w3.org

Pour
public-lld <public-lld@w3.org>
Copie

Objet
Re: Question about MARCXML to Models transformation



On Tue, 8 March, Ross wrote:
> This is not to say that the FRBR model is wrong or even necessarily 
flawed.
> I just think that applying it verbatim to RDF through OWL with an
> application profile that is intended to enforce its rules is more likely 
a
> barrier to adoption than it is insurance of semantic interoperability.

On Tue, 8 March, Jeff wrote:
> The constraints found in OWL could be enforced by another layer such as 
> Pellet ICV or Application Profiles, but we shouldn't assume these layers 

> are implied in the "strictness of FRBRer".

On Tue, Mar 8, 2011 at 4:06 PM, Richard Light <richard@light.demon.co.uk> 
wrote:
> I strongly agree with the thought that an entity can be given a URL, and
> thereby you can finesse the need for the "concept is the sum of its
> properties" approach. We will have many similar cases in the museum 
world,
> where information about an entity of interest (person, place, event, 
....)
> will be incomplete, or uncertain, or both. This shouldn't stop us from
> asserting what we _do_ know (or believe).

To summarize, can we say the following?

FRBR and RDA can improve the precision of resource description
and increase the opportunities for sharing descriptions at
various levels by making modeling distinctions grounded in
a coherent intellectual model.

However, for the linked data context, outside of the library
silo -- where knowledge about the things being described may
be imperfect, where the people making descriptions may have
an imperfect grasp of the models or of their applicability,
and where people may have data or software that lack clear
support of the models -- FRBR and RDA should be made available
for use in a form that is ontologically tolerant.

The sort of strict enforcement of rules and that served the
cause of data sharing in a time when data exchange required
the integrity of shared formats is not only not necessary
in the more loosely aligned linked data context - it is
counterproductive.

The FRBR and RDA vocabularies can be defined in an
ontologically tolerant manner, such that data which uses the
models imperfectly -- or data about things to which the models
imperfectly apply -- will not raise fatal exceptions when
linked with data that may be simpler, vaguer, or simply based
on different models.  Apparent misalignments, or contradictions
to the logic of the models, or gaps in descriptions, should
be flagged with nothing stronger than helpful error messages.

Application profiles, whether defined using OWL constraints
or through other means, still provide a way to constrain the use
of such vocabularies to an arbitrary degree of strictness
for the purposes of enforcing data integrity within a silo.

Hard-coding such constraints into the vocabularies themselves
imposes that ontological strictness on all downstream users
of the vocabularies, thus raising the bar to their adoption
and compromising their potential impact outside of the
library world.

Tom


-- 
Tom Baker <tbaker@tbaker.de>





Exposition  Visions d'Égypte. Émile Prisse d'Avennes (1807-1879)  - du 1 er  mars 2011 au 5 juin 2011 - BnF - Richelieu / Galerie Mansart  Avant d'imprimer, pensez à l'environnement. 
Received on Wednesday, 9 March 2011 15:21:57 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 March 2011 15:21:58 GMT