W3C home > Mailing lists > Public > public-lld@w3.org > November 2010

Re: frad:Person and foaf:Person

From: Karen Coyle <kcoyle@kcoyle.net>
Date: Mon, 01 Nov 2010 06:57:26 -0700
Message-ID: <20101101065726.bel08yri84k4kccc@kcoyle.net>
To: Richard Light <richard@light.demon.co.uk>
Cc: Dan Brickley <danbri@danbri.org>, public-lld@w3.org
Quoting Richard Light <richard@light.demon.co.uk>:


>
> I don't understand the reasoning (on page 7) behind putting Name and
> Identifier into a single box. Is that saying that each Name has a
> corresponding Identifier?  To me, names and identifiers are independent
> of each other, and have different properties and purposes.

I believe the box is there because they are both identifiers -- one  
uses the library practice of an authoritative form of the name, the  
other is more like what SemWeb considers an identifier, preferably a  
URI. In an earlier version of the FRAD documentation that I have, the  
definitions make clear that identifiers are assigned while names come  
with the bibliographic entity. Authoritative names are designed to  
both identify and display to human readers. (This makes the connection  
between identifier and controlled access point a bit tenuous, in my  
mind, because the access point is what is carried in the bibliographic  
record, and AFAIK is a human-readable string. But that's library  
arcana.)

What also puzzles me is the difference in the connectors between name  
and identifier. It seems to say that a name can be associated with  
more than one bibliographic entity, but an identifier cannot. Is that  
because FrAD recognizes that there are names that are ambiguous and  
are not (yet?) associated with only a single bibliographic entity?

Note that FRAD models Persons as a type of Bibliographic Entity, even  
though the definitions I have seen in earlier FRAD documentation  
define Person in much the same was as foaf. I guess this works, but it  
seems to me that logically FRAD:Person would then be a subclass of  
foaf:Person because it is narrower in scope. FRAD is uninterested in  
non-bibliographic aspects of a Person (that their hobby is gardening  
or taking long walks on the beach) and is also uninterested in Persons  
who will never create nor be the subject of a bibliographic object.  
Actually, it might be better to say that FRAD is uninterested in  
Persons *until* they create, etc. And in terms of data, traditional  
library authority data essentially stops at the name, while foaf  
gathers data from a broader environment in which the Person may  
operate. Library name authority information has been terribly narrow  
and doesn't facilitate inferencing from, say, email addresses and web  
sites to identities.

  Creating an identity for an author,
> e.g. by adding birth and death dates to their name, is a slightly
> artificial exercise with exactly the same purpose as minting a LD URL
> for that author.

Exactly.

>  Two differences between this technique and the LD
> approach are that:
>
>  - it won't scale up to identify all people uniquely
>  - it's not dereferencable

Also:

   - When it becomes necessary to change the display form, the  
identifier also changes (yes, this happens with some frequency,  
although more for subject headings than persons)

Very disruptive of existing data.

>
>> There is no 'things in the world' concept in library cataloging in the
>> sense that there is in SemWeb. This is in part because the library
>> catalog is a closed environment where all references are to other
>> things in the library catalog (or potentially in the library catalog).
>> Creating a mind meld between this model and the SemWeb model is going
>> to take some fancy footwork.
>
> Why try?  The current VIAF approach for authors, as I understand it, is
> to publish the bibliographic authority data, and then alongside it
> publish FOAF etc. views of the same person.  At least that's what I
> think I get when I ask for:
>
> http://viaf.org/viaf/102333412/rdf.xml
>
> [currently down for maintenance].  It certainly isn't a good idea to
> compromise your core data model.

Yes, of course. What I'm wondering about, though, are the semantics of  
that and where and how we make the connections. All we have in library  
data are various forms of the name (if there is more than one possible  
name form). In library data, within a given authority file each name  
is unique. In foaf, there is no control over names -- they are  
informative but not normative. I see no reason NOT to make the  
connection between library authority data and foaf, but I guess I've  
already jumped to the next step: how do we do something useful with  
that?

kc


>
> Richard
> -- 
> Richard Light



-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
Received on Monday, 1 November 2010 13:58:16 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 1 November 2010 13:58:17 GMT