- From: Ray Denenberg <rden@loc.gov>
- Date: Wed, 14 Nov 2001 10:23:00 -0500
- To: ZIG <www-zig@w3.org>
There was a proposal presented at the October ZIG meeting to change the holdings schema. See: http://lcweb.loc.gov/z3950/agency/zig/meetings/uk2001/holdings.html There was objection to the proposal, and we ended the meeting with no clear path towards resolution. There was agreement at the meeting to discuss this over the list, but there has been little discussion and no progress towards resolution (as far as I can tell). I've recently been asked (privately) to see if we can move the discussion along towards consensus. I would like to begin by trying to describe the problem in my own words, partly to get a better understanding myself. BibPart may have "child" bibParts, and this is represented by recursion, that is, Bibpart includes an element childBibPart whose data type is BibPart. Bibpart in addition includes elements enumeration and chronology; these two elements would occur within the child bibparts, as well as the top level bib part. Enumeration and chronology occur with each bibPart, and they too are viewed as hierarchical, for example in the enumeration "volume 5, Issue 2", "Issue 2" is subordinate to "volume 5". It is claimed that the schema cannot express "volume 5, issue 2" as a flat value. (Why it can't isn't clear to me. There doesn't seem to be any restriction on the value, but let's assume you can't do it, for argument sake.) So the suggestion is that enumeration and chronology each be defined as individually recursive, to allow children, thus to express the subordinate relation. Thus every bibPart (top level and children) would have a recursively defined enumeration and a recursively defined chronology. Those who oppose the proposal suggest that the recursive definition of bibPart is sufficient recursion to recurse these two elements (implicit recursion). In other word, suppose there aren't really any child bib parts, but there are "child" chronologies for the single bib part. You could artificially recurse bibpart to effect the recursion. That's my summary of the proposal and the two positions. Is this a reasonable interpretation? If so, I have two observations/opinions: 1. I don't think that artificial recursion of bibPart (i.e. implicit recursion) is a good thing. You shouldn't recurse bibpart unless there is a child bibpart. If you do, you have a semantic mess. Suppose we adopt these semantic and there is a child bibpart: how would you know whether the recurring enumeration/chronology applies to the child or the parent? 2. On the other hand, it seems like overkill to recurse enumeration and/or chronology. Why can't they simply be made repeatable? I.e. allow multiple occurences, where the semantics of multiple occurences is that the N+1th occurence is subordinate to the Nth. Comments please! --Ray -- Ray Denenberg Library of Congress rden@loc.gov 202-707-5795
Received on Wednesday, 14 November 2001 10:22:59 UTC