‘Access Points’ as Identifiers

An essay I originally posted to rda-l on 14 February 2007, and put here now mainly to have a persistent URL to easily access it. I made a few minor edits for clarity while I was at it. (So this is perhaps a new Expression of my essay, if you’re keeping track).

“Access points” as “Textual identifiers” ?

I have thought for a while now that RDA needs a better, more rigorous, clear, elegant and simple conceptual grounding of what are currently being called ‘controlled access points’ (or sometimes just ‘access points’, which sometimes means the same thing and sometimes not; previously ‘headings’, ‘citations’, ‘main/added entries’, etc.).

At the CC:DA meeting in Seattle, I heard someone (Jennifer Bowen?) talking about “access points” vs. “identifiers”, and how RDA would like to make identifiers optional or even preferred for linking records instead of ‘access points’, but it wasn’t clear how that would be done.

This got me started thinking, and lead to a line of thought, influenced very significantly by Martha Yee’s frequent use of the term/concept ‘work identifier’, that I explicate here in this draft essay unfortunately more lengthy than I wish it were.

My proposal is that our traditional system of controlled ‘headings’, ‘access points’, or ‘main entries’, really _is_ a system of identifiers, and best understood and discussed that way. A traditional heading is best understood as intending to uniquely and unambiguously identify a bibliographic entity—in the FRBR regime, a FRBR entity of some kind. A work, manifestation, item, person, corporate body, subject, etc. Constructing a heading (or ‘controlled access point’ in the not always consistent and somewhat confusing terminology of current RDA) is really constructing an identifier for one of these entities, within a particular system that has traditionally been the primary system of identifiers in the catalog. For lack of a better term, let us call identifiers constructued according to one version or another of this traditional system of headings: “Textual Identifiers”.[1]

Uses, Not Kinds

RDA terminology behind “controlled access points”, discussing “primary access points” vs “secondary access points”, implies that there are two KINDS of “controlled access points.” This is inaccurate and misleading terminology. In fact, there are instead two _uses_ of textual identifiers, or two tasks involving textual identifiers that the cataloger has to face in creating a record:

1) ESTABLISHING ‘textual identifiers’ for newly established FRBR entities. This includes, of course, establishing a textual identifier for the record being cataloged (a manifestation). This use or task roughly corresponds to the concept of ‘main entry’. This also includes establishing identifiers for People, Corporate Bodies, perhaps Subjects, perhaps Works and Expressions if cataloging codes are to require such (as Bowen suggested RDA was inclined to do).

2) REFERENCING the textual identifiers of other entities in the record for the entity at hand, in order to establish RELATIONSHIPS. These textual identifiers of other (‘foreign’) entities may have already existed before the moment of cataloging, or may have been ESTABLISHED _in order to_ allow their referencing to encode a relationship. This corresponds very closely to ‘added entry’–we will understand the ‘added entry’ as a _use_ of _another entity’s_ textual identifier, in order to establish a relationship between the entity at hand and that foreign entity. This is in fact also quite what happens with 6xx subjects (referencing a ‘textual identifier’ for a topic or other subject entity or entities), and with most other data elements you can think of which REFERENCE a ‘textual identifier’ (that is, a ‘heading’, or ‘controlled access point’) for another entity, in order to encode a relationship between the entity at hand and that other (‘foreign’) entity.

Traditional Practices Described In This Framework

Using this new terminology, how do we translate or explain certain cataloging practices?

“Title Main Entry” => “The textual identifier for this Group 1 entity is constructed using only the entity’s Title element.”

“Author Main Entry” => “The textual identifier for this Group 1 entity is constructed using BOTH the Title element, and the Author.” Actually, to be clear. “… the Title element and the Textual Identifier for the author, or one of the authors when multiple authors exist, as chosen according to the code.” It should be clear that ‘author main entry’ really means the textual identifier consists of the Author AND the Title, not _just_ the author. [This is a difference between the ‘main entry’ terminology, which pre-dates even the card catalog, and really makes sense only in the environment of the bound catalog(!) –And the ‘textual identifier’ terminology more suited for the digital era.]

It also becomes clear that a MARC 100, for instance, in current cataloging, in fact serves at least TWO purposes–it is one component of the Textual Identifier ESTABLISHED for the manifestation at hand (the one described by the bib record). AND, it REFERENCES the textual identifier of a Person entity, to describe a relationship between the manifestation at hand and that Person. An author 700, on the other hand, plays no part in ESTABLISHING the textual identifier for the manifestation at hand, but only describes a relationship. An author- title 700 likewise simply REFERENCES a textual identifier–in this case of another group 1 entity–in order to establish a relationship.

But these are not different ‘kinds’ of identifiers, they are instead different uses. Using a person identifier as part of the construction (establishment) of a work (and/or manifestation) identifier; vs. using a person identifier solely to record a relationship (referencing to record relationship).

This conceptual framework and terminology also points out some ambiguities in our current practice. What is a uniform title, or ‘uniform title main entry’? Does a string in a MARC 240 ESTABLISH a textual identifier for the Manifestation described in the bib record, OR for an Expression OR for a Work? Or does it somehow do multiple duties for all of these? We really aren’t neccesarily sure from current bib records and practice. Does it REFERENCE the identifier of another Work or Expression? Almost certainly, but which? Well, deciding/understanding that depends as a pre-requisite on understanding the establishment of textual identifiers for all entities involved.

A new way to describe cataloger’s tasks?

This terminology provides a more rigorous and clear conceptual framework for what’s going on, that not only will be easier to understand for ‘non library communities’, but will allow the kind of principle based foundation that RDA aspires to, making it more clear what the point is. This framework suggests two tasks of cataloging practice:

1) Whenever a new record for any FRBR entity is created, the cataloger must determine and establish the ‘textual identifier’ for that record/entity. (To allow other records to encode relationships to this record by traditional ‘textual identifier’ system.) [And our records must be structured such that this textual identifier is explicit and machine-addressable!] [Even if a manifestation bib record is traditional ‘uniform title main entry’, you still need to establish a textual identfier for the manifestation as distinct from a textual identifier for the work or expression.]

2) Whenever a cataloger references a textual identifier for a ‘foreign’ entity—that is, enters a ‘controlled access point’ for some entity other than the one she is writing the record for–the _purpose_ of this action is in fact to establish a relationship between entities. The cataloger must first explicitly decide precisely which entity she means to reference (the ambiguity of current uniform titles is highly undesirable), and secondly discover the already established Textual Identifier for that entity by consulting a bib or authority record, or by instead establishing the textual identifier for that other entity. (And ideally by establishing a corresponding bib or authority record to record that decision, but this isn’t always the way our current cataloging environment works).

This makes it clear how to talk about our traditional textual identifiers vs. ‘new fangled’ identifiers too. You can establish identifiers from any other system of identifiers (some type of URI or URN perhaps) in #1 along with our traditional textual identifiers; and you can reference these ‘new fangled’ identifiers in #2 instead of our traditional textual identifiers.

Incorporation in RDA

It is my suggestion that this terminology and framework ought to be at the heart of RDA, especially part B. Of course, this conceptual frameowrk doesn’t entirely match the way things are currently done or described. It may or may not be feasible to suggest changing the way things are done to fit into this more rigorous and clear conceptual framework—but it would still be valuable for RDA to re-organize it’s treatment baed on this framework, and to be clear about when legacy practices are kept _even though_ they do not match RDA’s conceptual framework.

But even to go this far would require significant re-thinking, re- organizing, and re-writing, beyond which seems to be contemplated by the very short schedule for RDA Part B and the continued assurances that this is because Part B will be “unchanged from AACR2”. This way of thinking about headings is not particularly new or original to me; as I mentioned it is something Martha Yee (and others) have been approaching for years, and I think it is in fact the way the JSC and Editor are thinking about things. But it does not seem like it will make it into RDA in a real way. This worries me. I think these issues of controlled textual identifiers as relationship indicators may be more important for bringing cataloging practice into the digital era than the minutia of transcription decisions which much of Part A consists of.

I appreciate any comments, feedback, debate, disagremeent on these ideas.

Jonathan Rochkind

[1] I realize it could be argued that many other kinds of identifiers are ‘textual’, depending on what is meant by ‘textual’, perhaps including URIs and such. The intention of defining the term ‘textual identifier’ as I have is not to argue this point. But we need a term that refers explicitly to headings constructed according to our traditional systems, and makes it clear that they are to be understood as one type of identifier. ‘Textual identifier’ will do as a term of art, unless someone can come up with a better one.

3 thoughts on “‘Access Points’ as Identifiers

  1. Jonathan, I agree with your analysis of the muddle surrounding the ‘controlled access points’. When I read the new intro to RDA I was struck by how hard it would be to make sense of if you didn’t know the historical development of the bib record and library catalogues.

    On the ‘new-fangled’ identifiers: has much work been done to your knowledge toward developing non-textual identifiers for frbr entities? If an identifier is ‘a label within a namespace’, then we presumably need to decide what the namespace(s) is/are going to be and decide on a system for maintaining them. And if there are multiple namespaces, then some way of mapping between identifiers in each for the same entity. (http://oclc.work.100008 = http://lc.work.2346). One advantage of textual identifiers is that cataloguers in different places in the world can (theoretically) come up with the same identifier for a work if they’re following the same rules.

  2. I’m not sure what work has gone on, no. I get the impression that not too much has gone on, and now the time is right for it to start. On one of the listservs, someone talked about an effort to create international control numbers for authority records (is that part of the VIAF?), and how that was generally considered a dead-in-the-water prospect at this point.

    I think that it’s worth a start to just apply to a metadata/URI type regime to one ‘universe’ of authority data (say, the US-centric NAF/SAF), rather than try to consolidate things internationally right away. That would still be valuable.

    Because, after all, even though it is an interesting feature of the current textual identifier regime that catalogers in different places working independently with no coordination can “theoretically” come up with the same identifier—in practice, how important is this? In the US, a cataloger that created a name authority heading without consulting the NAF would be acting in extreme folly! No? While theoretically the system is set up to support people working without coordination, we don’t actually do that!

    But I’m not sure I’d advocate getting rid of our traditional system of ‘textual identifiers’ anyway. I DO advocate experimenting with new systems of URI-type identifiers. But more importantly, if we are to keep it, we need to rationalize it as a system of identifiers. For every FRBR entity we are controlling, we need to know it’s ‘textual identifier’. For every ‘textual identifier’ used in a record, the cataloger needs to be clear on precisely what entity’s identifier that is, what relationship to what entity is being recorded, or what entity is having it’s textual identifier established. That is not generally the way we approach things right now (in part because, before the FRBR model, we didn’t always have the language to describe this clearly), and it leads to problematic ambiguity. An ambiguous identifier is not a very good identifier at all.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s