Received: by nummer-3.proteosys id <01C19443.4DD3E014@nummer-3.proteosys>; Thu, 3 Jan 2002 11:42:13 +0100 Return-Path: <@vm.gmd.de:LATEX-L@DHDURZ1.BITNET> MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C19443.4DD3E014" x-vm-v5-data: ([nil nil nil nil nil nil nil nil nil][nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil]) X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message Subject: References and citations (3) Date: Sat, 9 Nov 1991 22:40:38 +0100 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: From: Sender: "LaTeX-L Mailing list" To: "Rainer M. Schoepf" Reply-To: "LaTeX-L Mailing list" Status: R X-Status: X-Keywords: X-UID: 454 This is a multi-part message in MIME format. ------_=_NextPart_001_01C19443.4DD3E014 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable You may notice the "In response to some similar stuff, Nico Poppelier remarked ... " note in the margin of section 5.5.6 of my notes. You'll find copies of the relevant correspondence with Nico appended below. = I'll leave it to you to think through the implications. (? ?) -------------------------------------------------------------------------= ---- Subject: Do you know of any good SGML treatments of = citations/reference-lists? Date: 6 Oct 91 19:58:15 To: n.poppelier@nl.elsevier I've been looking at some SGML documents: The AAP publications about electronic manuscripts and markup. (These = are the ones you told me about, as obtained from EPSIG, Dublin, Ohio.) The Text Encoding Initiative's draft "guidelines for the the encoding = and interchange of machine-readable texts". On the one hand, they seem to provide some ready-made analyses of = structure that could usefully be stolen for LaTeX 3 (e.g., the structure of front matter). On the other hand, I was disappointed that neither of them seems to provide a ready-made analysis of the variety of forms of citation and reference-list (and how one can cope with this variety). E.g.: * the author-date convention is that "if the author occurs naturally in = a sentence, you just give the date", and this is catered for by .sty files in TeX archives that give a choice between \cite and \shortcite. (One can quibble about other aspects of these .sty files, but at least = they exist now.) The "short form" scheme as often used in footnotes doesn't = seem to get much consideration either. The SGML documents don't seem to = address the subtleties, and its not clear what an author is supposed to do. (In fact, if the example on page 8 of the AAP reference-manual is done = in an author-date style, it would end up as "Tobias states ... (Tobias, = 1980) ... " at which a human copy-editor would presumably be horrified, and would change it manually to "Tobias (1980) states ... ". So much for = SGML enabling everything to be automated!) * there are potentially a lot more document-types that may need to = appear in reference-lists than are considered by either SGML document. E.g. neither seems to deal with the "in conference-proceedings" type of = thing (unless I've not read them carefully enough to interpret them properly = for such things). Have you come across any better attempts by SGML-ers to define DTDs that deal properly with citations and reference-lists? I suppose I'm looking for an SGML analogue of ISO standard 690. A supplementary AAP report on the subject, perhaps (like their supplementary report about markup of complex tables)? I ask partly to see whether there is any off-the-shelf analysis that could be stolen for LaTeX 3 partly because I'm surprised at the superficiality of the approach taken = in the AAP documents, given that there are documents like ISO 690 that give ideas about what needs catering for. (Its not an easy area, = but I'd have thought they would tried to provide deeper coverage than they = have.) partly because the TEI has made preliminary enquiries about whether I = want to attend a working party on these and other matters. (I'm not sure = that I'm really suitable, since I'm not all that fluent in SGML!) (If nothing more comprehensive exists in the SGML world, maybe I should write to the AAP drawing their attention to the apparent gaps in their coverage.) David Rhead d.rhead@uk.ac.nottingham.ccc.vme -------------------------------------------------------------------------= ---- Subject: RE: Do you know of any good SGML treatments of citations/reference-lists? From: N.POPPELIER@nl.ELSEVIER Date: Wed, 9 Oct 1991 11:27:58 +0000 To: David_Rhead@uk.ac.nottingham.ccc.vme (Receipt Notification = Requested) (Non Receipt Notification Requested) David, I've read the criticisms in your message "Do you know of any ..." and will try to answer them. First of all, you make a mistake that SGML novices often make: SGML is not the solution to all sorts of problems. For any class of documents you handle, you will have to make a structural analysis: which objects can occur in these documents, and what are their relationships? Strictly speaking, you don't allow presentation matters to creep into this analysis. The language for expressing the structured document analysis, or definition, is SGML. It's the same with programming: suppose you have a numerical problem that you need to solve by computer. You write it down in the language of mathematics, and then in a programming language. The programming language is the vehicle, but _you_ have to come up with the solution. Now, I agree with you that none of the existing dtd's contain modules for references and citations that are sufficient for author's and publisher's purposes. I sometimes think that a translation of the structure of BibTeX's standard bibliography styles into SGML is the most sophisticated of everything existing so far. Another thing is that the difference between \cite and \shortcite is not a matter of document structure, but one of document semantics. If the author name occurs naturally in a sentence, you just give the date. That's semantics, and dtd's have nothing to do with this, unless you write down the syntax of English sentences in SGML, and use the fact that an author name is present to suppress it in a citation in that same sentence. Even then, you will have to compare these names, which again is semantics! You see, it has nothing to do with structure. What you will have to do is define an element in your dtd, say for reference to bibliography, that has at least 1 attribute, say type, that can have values "long" and "short". That is, you give the person who writes a document according to this dtd the means to express the fact that a certain author name must be suppressed. I'm not aware of existing analyses of the citation-reference matter, apart from what the TEI Draft Report says on this. Last week there was a meeting on SGML in physics publishing, called together by the European Physical Society, and the publishers who were "into SGML" agreed that existing dtd's are insufficient. Before the end of this year we hope to finish a structural analysis of our scientific journals, including front matter, references, citations, tables, math, indexes, ... That will serve as input to the various committees that write or rewrite dtd's. And there are more of these committees than you think! All of them tend to ignore the things I mentioned at the start of this paragraph. ------_=_NextPart_001_01C19443.4DD3E014 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable References and citations (3)

You may notice the "In response to some similar = stuff, Nico Poppelier
remarked ...  " note in the margin of = section 5.5.6 of my notes.  You'll
find copies of the relevant correspondence with Nico = appended below.  I'll
leave it to you to think through the = implications.  (<cite.ad type=3Ddate>?
<cite.short type=3Dfirst>?)

----------------------------------------------------------------= -------------
Subject:  Do you know of any good SGML = treatments of citations/reference-lists?
Date:     6 Oct 91 = 19:58:15
To:       = n.poppelier@nl.elsevier

I've been looking at some SGML documents:
The AAP publications about electronic manuscripts and = markup.  (These are
    the ones you told me about, as = obtained from EPSIG, Dublin, Ohio.)
The Text Encoding Initiative's draft "guidelines = for the the encoding and
    interchange of machine-readable = texts".

On the one hand, they seem to provide some ready-made = analyses of structure
that could usefully be stolen for LaTeX 3 (e.g., the = structure of front
matter).

On the other hand, I was disappointed that neither of = them seems to
provide a ready-made analysis of the variety of forms = of citation and
reference-list (and how one can cope with this = variety).  E.g.:
* the author-date convention is that "if the = author occurs naturally in a
  sentence, you just give the date", and = this is catered for by .sty
  files in TeX archives that give a choice = between \cite and \shortcite.
  (One can quibble about other aspects of these = .sty files, but at least they
  exist now.) The "short form" scheme = as often used in footnotes doesn't seem
  to get much consideration either.  The = SGML documents don't seem to address
  the subtleties, and its not clear what an = author is supposed to do.
  (In fact, if the example on page 8 of the AAP = reference-manual is done in
  an author-date style, it would end up as = "Tobias states ... (Tobias, 1980)
  ...  " at which a human copy-editor = would presumably be horrified, and
  would change it manually to "Tobias = (1980) states ...  ".  So much for SGML
  enabling everything to be automated!)
* there are potentially a lot more document-types = that may need to appear
  in reference-lists than are considered by = either SGML document.  E.g.
  neither seems to deal with the "in = conference-proceedings" type of thing
  (unless I've not read them carefully enough to = interpret them properly for
  such things).

Have you come across any better attempts by SGML-ers = to define DTDs that
deal properly with citations and = reference-lists?  I suppose I'm looking
for an SGML analogue of ISO standard 690.  A = supplementary AAP report on
the subject, perhaps (like their supplementary report = about markup of
complex tables)?  I ask
partly to see whether there is any off-the-shelf = analysis that could
     be stolen for LaTeX 3
partly because I'm surprised at the superficiality of = the approach taken in
     the AAP documents, given = that there are documents like ISO 690 that
     give ideas about what needs = catering for.  (Its not an easy area, but I'd
     have thought they would = tried to provide deeper coverage than they have.)
partly because the TEI has made preliminary enquiries = about whether I want to
     attend a working party on = these and other matters.  (I'm not sure that
     I'm really suitable, since = I'm not all that fluent in SGML!)
(If nothing more comprehensive exists in the SGML = world, maybe I should
write to the AAP drawing their attention to the = apparent gaps in their
coverage.)


David Rhead
d.rhead@uk.ac.nottingham.ccc.vme
----------------------------------------------------------------= -------------
Subject:  RE: Do you know of any good SGML = treatments of
          &nbs= p;        = citations/reference-lists?
From:     = N.POPPELIER@nl.ELSEVIER
Date:     Wed, 9 Oct 1991 = 11:27:58 +0000
To:       = David_Rhead@uk.ac.nottingham.ccc.vme (Receipt Notification = Requested)
 (Non Receipt Notification Requested)

David,

I've read the criticisms in your message "Do you = know of any ..." and
will try to answer them.

First of all, you make a mistake that SGML novices = often make: SGML is
not the solution to all sorts of problems. For any = class of documents
you handle, you will have to make a structural = analysis: which objects
can occur in these documents, and what are their = relationships?
Strictly speaking, you don't allow presentation = matters to creep into
this analysis.

The language for expressing the structured document = analysis, or
definition, is SGML. It's the same with programming: = suppose you have a
numerical problem that you need to solve by computer. = You write it down
in the language of mathematics, and then in a = programming language. The
programming language is the vehicle, but _you_ have = to come up with the
solution.

Now, I agree with you that none of the existing dtd's = contain modules
for references and citations that are sufficient for = author's and
publisher's purposes. I sometimes think that a = translation of the
structure of BibTeX's standard bibliography styles = into SGML is the
most sophisticated of everything existing so = far.

Another thing is that the difference between \cite and = \shortcite is
not a matter of document structure, but one of = document semantics. If
the author name occurs naturally in a sentence, you = just give the date.
That's semantics, and dtd's have nothing to do with = this, unless you
write down the syntax of English sentences in SGML, = and use the fact
that an author name is present to suppress it in a = citation in that
same sentence. Even then, you will have to compare = these names, which
again is semantics! You see, it has nothing to do = with structure.

What you will have to do is define an element in your = dtd, say <rb> for
reference to bibliography, that has at least 1 = attribute, say type,
that can have values "long" and = "short". That is, you give the person
who writes a document according to this dtd the means = to express the
fact that a certain author name must be = suppressed.

I'm not aware of existing analyses of the = citation-reference matter,
apart from what the TEI Draft Report says on this. = Last week there was
a meeting on SGML in physics publishing, called = together by the
European Physical Society, and the publishers who = were "into SGML"
agreed that existing dtd's are insufficient.

Before the end of this year we hope to finish a = structural analysis of
our scientific journals, including front matter, = references, citations,
tables, math, indexes, ... That will serve as input = to the various
committees that write or rewrite dtd's. And there are = more of these
committees than you think! All of them tend to ignore = the things I
mentioned at the start of this paragraph.

------_=_NextPart_001_01C19443.4DD3E014--