See, the tricky thing is, that our existing 'access points' were used 
for TWO purposes, both identifying an entry for humans _and_ collocating 
related entries. (At least when the 'access points' are what we somehow 
used to call 'controlled headings'--what I _think_ RDA refers to as 
'preferred access points'. RDA's terminology is not helpful in clearing 
up confusion here).

That traditional 'collocating' function is pretty much completely 
analogous--from a pre-computer world--to the function of machine 
identifiers.   The properties that made these 'access points' suitable 
for collocating related entries are the same properties that make them 
suitable as machine identifiers.  (more or less, with some significant 
issues, but better than anything else we've got in our existing data and 
practices).

Yes, that we use the same device for both purposes does lead to some 
issues, but it's the legacy we inherit. It would be of value to separate 
these functions, but that doesn't seem in the cards at the moment.

Jonathan

Alistair Miles wrote:
> Hi Karen,
>
> Just a quick comment, I understood that "access points" were means of
> identifying an entity for humans. Isn't this separate from identifiers
> which are for use within software, which need not ever see the light
> of day outside software?
>
> Coming up with an author/title "access point" for a work or expression
> sounds like something quite different from coining a persistent URI to
> identify the same work or expression for use within web-based
> information systems.
>
> I understand that LC are creating "permalinks" for bibliographic
> records? I think the "permalink" idea is a great way to tackle the
> need for identifiers. I.e. if each authority created a permalink URI
> identifying each bibliographic entity for which they hold a record, we
> could then use the permalink URIs in our RDF/RDA metadata to link all
> the entities together.
>
> Cheers,
>
> Alistair
>
> On Mon, Dec 08, 2008 at 06:00:17PM -0800, Karen Coyle wrote:
>   
>> A recent discussion on the RDA-L list brought to light some
>> information about FRBR relationships and how RDA creates (or does not
>> create) identifiers for linking between work, expression,
>> manifestation, and item. (WEMI)
>>
>> The key information is in chapter 17, which is only 17 pages long (and
>> a lot of it is examples). What is significant for this project is that
>> links between WEMI consist of "preferred access points" that represent
>> the FRBR entity. For example, there can be an author/title access
>> point that represents the work ("Schumann, Clara, 1819-1896. Scherzos,
>> piano, no. 1, op. 10, D minor"). A expression access point represents
>> the expression of the work ("Blade runner (Motion picture : Final
>> cut"). Some works and expressions have LC authority records, and the
>> ID number of the authority record may be considered an identifier for
>> the FRBR entity.
>>
>> The chapter also shows examples using what I see as "external"
>> identifiers for the FRBR Group1 entities (WEMI). These include the
>> ISBN or music publisher numbers for manifestations, and the
>> International Standard Text Code for works. These external identifiers
>> strike me as problematic for a few reasons:
>>  - coverage is uneven: there are many works and manifestations that
>> don't have such a code
>>  - coverage is uneven: there are no such codes for expressions
>>  - these identifiers come from another information space, and creators
>> of RDA data cannot create or correct them when needed
>>
>> The upshot is that identifying and linking FRBR Group1 entities using
>> these external identifiers is spotty at best, and decidedly not
>> consistent enough to create reliable FRBR relationships.
>>
>> Now for the catch:
>>
>> RDA does not specify "preferred access points" for manifestations or
>> items. So we have no "identifier" for manifestations. (not having one
>> for items is a bit less of an issue, for various reasons). So we can
>> link from:
>>
>> manifestation (with an expression access point) to an expression (with
>> a work access point) to a work
>>
>> But we can't link from an item to a manifestation, nor can we create
>> relationships between manifestations. Well, not with what we have
>> today in RDA.
>>
>> This came up for us in Alistair's analysis of scenario 2, where he
>> used the ISBN for the manifestation identifier. I objected, but in
>> fact there is no actual manifestation identifier to use.
>>
>> I don't really know what to do at this point, but it would be good to
>> try to make use of RDA access points as WEMI identifiers in one or
>> more of the scenarios so we can illustrate the issue. I'll see what I
>> can do, but may need help.
>>
>> kc
>>
>> -- 
>> --  ---
>> Karen Coyle / Digital Library Consultant
>> [log in to unmask] http://www.kcoyle.net
>> ph.: 510-540-7596   skype: kcoylenet
>> mo.: 510-435-8234
>> ------------------------------------
>>     
>
>   

-- 
Jonathan Rochkind
Digital Services Software Engineer
The Sheridan Libraries
Johns Hopkins University
410.516.8886 
rochkind (at) jhu.edu