The basic problem is that the names of the parts of the journal 
hierarchy don't tell you about their order. That's why the SICI standard 
(and some citation formats)  ignores what the parts are called and just 
gives the enumeration (yes, we can call it something else) as a 
highest-to-lowest list, like "4:3". Order makes a big difference in 
terms of actually identifying the part that you are citing ("3:4" is NE 
"4:3"). For journals using simple numbering, using volume/number works 
fine, but so many journals don't follow that pattern. If you have an 
issue that is called neither volume nor number, where do you put it? 
(Also, "volume/number" works only for English language journals; 
presumably you want to cite what the level is called on the cover of the 
journal so that a person looking at it can verify that they have the 
right item.)

Although my example code looks complex, it's not what I would expect 
anyone to have to code directly. Also, library systems can store 
patterns that pertain to individual journal titles, so the person keying 
can be presented with a pattern already set to be filled in based on the 
title or ISSN. For citation systems, it is probably enough for someone 
to fill in the levels and their "numbers" (which can sometimes be 
letters, months, seasons or dates) in the correct order, something like:

volume  1
part      B
number  6
date  [whatever]

And this would display nicely as: "volume 1, part B, number 6" or "v. 1, 
pt. B, no. 6" or "1(B)6" or any other number of citation formats. But if 
you haven't coded the order of the levels then things like linking (in 
the OpenURL sense) are likely to fail. So call them what you want, these 
are the data elements that are needed, and the relationship between them 
is key.


Bruce D'Arcus wrote:

>On 11/22/05, Karen Coyle <[log in to unmask]> wrote:
>>Bruce, I thought I had explained in my message why the method that you
>>suggest does not work. However, it might be better explained by someone
>>with some good, juicy examples from the reality of serials issuance. Anyone?
>Sorry, am rather distracted today with a ton of things, so no doubt
>missed some nuance.  I've also not come across any examples where
>"levels" are particularly meaningful in citations, so conveniently
>passed that over ;-)
>However, in any case, I do think we need to find a simpler example
>than your "simple" example.  One thing I have come around to over the
>past bit of time is that it's essential that our metadata models and
>formats be easier to work with for non-experts; both those who would
>create the metadata, and those who would write tools to consume it. 
>It's hard enough to convince people to even do decent DC.
>I realize my example is problamatic because I use the term "number"
>when we could be talking about, say, <ex:part>B</ex:part>.  But could
>you imagine achieving the same effect without having to use terms like
>"levels" or "enumerations"?  Even if different journals were to use
>"part" inconsistently, does it really matter, either for citations, or
>for finding the correct item?
>BTW, in RDF, one tends to takea a different tack than this approach of
>having generic metadata structures (say "enumeration"), with specific
>modifiers (a type attribute).  Instead, one would define specific
>structures, and then if you need to say that, for example, volume and
>issue are both kinds of enumerations, one would define that in the
>ontology (above the XML in effect).

Karen Coyle / Digital Library Consultant
[log in to unmask]
ph.: 510-540-7596
fx.: 510-848-3913
mo.: 510-435-8234