Re: Tim Berners-Lee on the Semantic Web

Andy's hit the nail on the head, this is the key point.

We absolutely NEED a formally and explicitly described domain model.  We 
didn't have one before FRBR. We had an implicit, not clearly defined, 
ambiguous, not always consistent, inherited shared mental model. 

I am confused by people saying they find it a 'minus' not a 'plus' that 
RDA is "based on FRBR".  Let's be clear -- the 'based on FRBR' that RDA 
is based on the _conceptual domain model_ outlined in FRBR. For better 
or worse, the actual 'user tasks' in FRBR are just a sideshow, RDA isn't 
really 'based' on them, and the FRBR domain model itself isn't really 
'based' on them. It's a fiction.  I am happier just ignoring the user 
tasks (regardless of how useful they are), because it's the domain model 
FRBR provides that is useful.

Simply expanding on AACR2 as rules for creating text is exactly the 
wrong approach. We don't need rules for creating text, we need rules for 
creating data elements in a defined domain model.  That defined domain 
model is what allows catalogers or metadata creators to use their 
"cataloger's judgement", understanding what they're doing. And is what 
defines data consumers to understand what the data they've got means 
without having to be catalogers, and to pull the data elements they want 
out of the data understanding what those data elements are intended to 
mean.  It's what allows us to create data that is flexible and will be 
useful in the future even for use cases we didn't think of previously, 
because we know what sort of data we were trying to create. Without such 
a formal domain model, you're just creating 'text', not 'data'.  Which 
is indeed what we used to do, when the text was destined for printed 
cards or pages.  But now we need data.

So if the FRBR data model isn't good (enough), you can expand on it -- 
or you can even abandon it entirely and create a new one. But it should 
be noted that also, for better or worse, the FRBR domain model was 
intended to be compatible with our legacy data and practices -- it is a 
formalization of cataloging tradition.   If you don't think there's a 
need to be compatible with legacy data, then, sure, abandon FRBR 
conceptual model and start from scratch maybe.  If you DO think there's 
a need to be compatible with legacy data, while also creating an actual 
explicit formal model of what we're doing, then I don't think you're 
going to do better by abandoning FRBR's model and starting from scratch.

But attempting to base itself on a formal domain model -- and in 
particular on the formal domain model the library created attempting to 
formalize our legacy practices -- is to my mind one of the RIGHT things 
that RDA did (or tried to do). Critisizing it for that seems very 
wrong-headed to me, as does trying to create an AACR2+ that still is 
just rules for writing text, without being built on a domain model.

Jonathan


Houghton,Andrew wrote:
>> From: Next generation catalogs for libraries
>> [mailto:NGC4LIB_at_LISTSERV.ND.EDU] On Behalf Of James Weinheimer
>> Sent: Tuesday, October 20, 2009 08:58 AM
>> To: NGC4LIB_at_LISTSERV.ND.EDU
>> Subject: Re: [NGC4LIB] Tim Berners-Lee on the Semantic Web
>>
>> I confess that RDA may be exactly what people want, but that remains to
>> be demonstrated, especially in the face of all of these studies of how
>> people use the web. The fact that it is based on FRBR only makes me 
>> more suspicious, not less.
>>     
>
> Personally, I believe that FRBR and RDA will be more useful to the 
> Semantic Web since they are both based on a domain model which can
> be expressed as an RDFS/OWL ontology.  People can argue whether the 
> domain model is correct, but having a domain model is important to
> understanding the data and its relationships within the domain model
> and outside the domain model, e.g., linked data.
>
>
> Andy.
>