Following in Roy's footsteps, I can provide a little ground-truthing
here.  I created a "roles" index for WorldCat Identities.  It is
browsable, as well as searchable, if you want to look at it.  When I
indexed it, I has a list of 192 codes and their full-text translations
and used both in my indexing.  That means I should see 384 index terms
in that index.  Instead, I see 1131 terms.

Here's a pointer into that index at the term "clr".  Replace the "clr"
in the URL with other starting places. Replace the term with an empty
string and you'll see the top of the index. The terms are hot and can be
clicked on to get to the Identities records themselves.

http://worldcat.org/identities/search/PersonalIdentities?operation=scan&
scanClause=%0Alocal.Role%20exact%20%22clr%22%0A&responsePosition=1&versi
on=1.1%0A%0A&maximumTerms=20

Ralph

-----Original Message-----
From: Bibliographic Framework Transition Initiative Forum
[mailto:[log in to unmask]] On Behalf Of Karen Coyle
Sent: Saturday, May 25, 2013 10:34 AM
To: [log in to unmask]
Subject: Re: [BIBFRAME] Role a nature of Bibframe authorities

On 5/25/13 6:47 AM, Kevin Ford wrote:
> > It is hard to fathom that role is being considered as part of an 
> > authority.
> -- It's not with the exception of one scenario, which I expect will be

> very, very few cases overall: when it is impossible to determine the 
> role because of poor cataloger entry in existing MARC records.

Given that there are tens of thousands of libraries using MARC in their
local systems, only some of which are visible outside their own systems,
I think it will be unwise to make decisions based on estimates of "very,
very few cases". In fact, world-wide, we don't know how many such cases
exist. Also, there are people using MARC whose language of cataloging is
not English, and therefore they use different sets of codes or terms for
roles (and for much else in the record).

Once again, I wish our focus were NOT on transitioning MARC at this
stage of the metadata development. I fear that we risk our future by
looking backward, not forward. Honestly, just throw the "bad" MARC
string into a "bad data" field and leave it in the bibliographic
description. It is NOT author/agent information, it is bibliographic
information, and should stay there.

kc


> This is the Bad Data example in the discussion paper.  I also 
> anticipate this will only ever be an issue during a transition phase, 
> meaning that, moving forward, specific "codes" or links will be used 
> to describe the relation of an authority entity to a work.  I also see

> the scenario as a limited accommodation to be made during said 
> transition phase.
>
> If $e says "editor" or "author of," we can associate those lexical 
> entries with relator codes.  If, however, $e says "edtor" or "autor 
> of," we cannot necessarily reliably associate those poorly entered 
> lexical entries with relator codes.  This is why it is a limited 
> accommodation during a transition period.  In the future, designating 
> a "role" would be done in a controlled manner.
>
> I'm not delighted about finding data entry errors in our current 
> bibliographic data, but I can see that they are a very small 
> percentage all told.  Is a more elaborate solution required for such a

> small amount of existing data, especially knowing we can improve on 
> this moving forward so that we do not have this problem?
>
> Cordially,
> Kevin
>
>
>
>
>
> On 05/24/2013 07:25 PM, J. McRee Elrod wrote:
>> It is hard to fathom that role is being considered as part of an 
>> authority.
>>
>> The person, family, or corporate body represented by an authority may

>> have any number of roles.  A person may be author, editor, 
>> illustrator, translator, depicted, or any other role to the work or 
>> instance listed in the RDA relator terms or MARC relator codes.
>>
>> There should be *one* authority per entity, and the relation(s) of 
>> that entity to the work or instance should be external to that 
>> authority, perhaps incorporated into the link?
>>
>> An entity may have more than one relation to a work or instance, 
>> e.g., actor/director, author/illustrator.  There should not be links 
>> to two or moore authorities for the same entity because of the two or

>> more roles.  There should be one access point per entity per 
>> work/instance, with role(s) expressed externally to the authority.
>>
>>
>>
>>
>>     __       __   J. McRee (Mac) Elrod ([log in to unmask])
>>    {__  |   /     Special Libraries Cataloguing HTTP://www.slc.bc.ca/
>>    ___} |__ 
>> \__________________________________________________________
>>

--
Karen Coyle
[log in to unmask] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet