[1148] in Public-Access_Computer_Systems_Forum
Re: Library Automation--Comments & Questions
daemon@ATHENA.MIT.EDU (Bernard Lang)
Fri Sep 4 10:58:26 1992
Date: Fri, 4 Sep 1992 09:52:44 CDT
Reply-To: Public-Access Computer Systems Forum <PACS-L%UHUPVM1.BITNET@ricevm1.rice.edu>
From: Bernard Lang <Bernard.Lang@inria.fr>
To: Multiple recipients of list PACS-L <PACS-L%UHUPVM1.BITNET@ricevm1.rice.edu>
----------------------------Original message----------------------------
> From: RSD@AppleLink.Apple.COM (Research SW Design, D Goldman,PRT)
> Subject: Re: Library Automation Questio
>
> Bernard,
> First, as the creator of one commercially-available bibliographic
> database program (PAPYRUS), I am hoping that you could expand on this
> part of your message:
>
> >I would also emphasize that a bibliographic database is most useful
> >when it can be directly used to incorporate citation in documents,
> >which is precisely the purpose of the BibTex/Refer format. However,
> >they become unmanageable when the database becomes large.
>
> Do you mean that the particular programs you have tried did not
> function well with large databases? Or do you feel that there are
> intrinsic problems with the idea of using _any_ personal bibliographic
> database program to maintain a large database? (And how large is
> "large"?)
First I am only a not too satisfied user who is trying to identify
his problems. I am not a librarian, nor an expert on this topic.
I have been using BibTeX while writing papers, like many of my
colleagues.
One of the problems with personnal bibliographic databases is
accuracy. Many people just don't care (look at published papers) and
occasionally pollute others to whom they give their database (which
one cannot really check. For example several parsing papers (including
some of mine) reference J. Kasami, whose first name happens to be
Tadao, because a mistake was made by one author 20 years ago, and the
original document is not easily available. Errors may also be due to
the limited competence of people who enter the information
(secretaries, or researchers (I often do not know what to make of
BibTeX entry structures).
The best solution I see is a global bibliographic database service,
that would be maintained and kept correct with the help of all users.
and would contain all the bibliographic info. All of it preferably on
internet.
Another problem is related to the way BibTeX is used. Each entry must
be indexed with a unique identifier that is used to cite it in a LaTeX
document. It is quite difficult to ensure the uniqueness, and the
identifier is not necessarily meaningful to many people (sharing the
database).
While unique identifiers can have their use (cf. object oriented
databases, and object identity), they should not have any meaning and
should be automatically generated by the database.
Instead, each user should be able, while writing a document and
within the document, to associate meaningful identifiers to database
entries, and use these identifiers in the document for citation
purposes. Note that I am talking of identifiers used by the authors,
but usually not visible in the final formatted document (I am not
talking WYSIWYG, but TeX style - but it could be adapted to WYSIWYG).
These identifier would then be local to the document and not included
in the database as is the case for BIbTeX.
> How large is "large"?
Well, not very large. The largest BibTeX base we have, shared by
several research group is a bit above 1 MB, which I consider small.
And it is already too big for BibTeX which it slows down considerably.
> Second, I hope that you can help me to understand what the problem is
> with:
>
> > ...a conference proceeding that is also an issue of 2 distinct journals,
> > all 3 being the same document. Or a PhD dissertation which is also
> > a technical report...
>
> If I follow you, you are talking about a situation in which a single
> document has been published more than once.
No, I mean a document published only once, the same cover indicating
that it is the proceedings of conference X in some series of
conferences, and also that it is the June issue of journal Y and the
July issue of Journal Z (e.g. ACM SIGPLAN Notices Vol.18 No.8, or
SIGPLAN Notices vol.26 No.4, or SIGARCH Computer Architecture News
Vol.17 No.2)
> How would you cite such a document in a _manuscript's bibliography_?
> I can see two likely methods:
> 1) You cite only the form that you had physically acquired and read,
> ignoring the others.
Not applicable in case above.
> 2) You cite one of the forms, using standard formatting rules, and
> then mention the others in a parenthetical note at the end of the
> citation.
Yes. This may necessary for the reader, if his librarian is careless,
or the library not computerized.
> Either of these methods can also be followed when using a
> bibliographic program, or when using a BibTeX or Refer type of format.
> One can also simply include the same reference three separate times,
> once for each published form, and allow the end-user to keep any or
> all of these for future use.
In any case, it is necessary to have the 3 forms in the database, and
it is also important to have links between them indicating that they
refer to the same publication. More generally many kinds of links are
essential or useful in a bibliographic database (and not available
with BibTeX):
- identity links as above
- includes/part-of links relating articles and books in which they
appear, or volumes to a multi-volume collection.
- filiation relating succesive versions of a paper (needs cooperation
and help of field specialists and users)
- probably many others.
Also, one also wants a separate base of authors, editors,
institutions, conference series, or journals etc... all related to the
bibliographic database. The reason is that their names can change over
time, while this is not necessarily known to the database user. For
example INRIA (my institution) used to be IRIA, Tim Teitelbaum first
published as Ray Teitelbaum, K. Vijay Shankar has changed his spelling
several times over the years, and (believe it or not) some women do
take their husbands' names. Also conferences in the same series have
changing names, often simply because the person in charge of the
proceedings does not bother to check the exact title.
What I mean is that author name should be given as appear in the
publication, while another field should contain a reference to this
author's entry in the author base. Again this requires some help from
specialists, since not everyone knows that 2 names mean the same
person, or that one name means two different authors: Thompson and
Ritchie (in that order) are well known for their work on Unix, and
another Thompson and Ritchie (in that order too) couple for their work
on parsing (though the first names differ). That is why I believe that
a proper database should be maintained on the net by concerned
specialists of the field (supervised by professional librarians).
To be honest, my concern here is bibliographic search, more than
help for bibliography insertion when writing papers. Still one issue
is that what information you have to ideally include in the
bibliography of a paper depends on the tools available to the readers
for bibliographic search.
I am pretty sure all this must be obvious to you specialists. But
BibTeX handles none of this, and the abbreviation system (e.g. for
conference names) gives only limited help (not usable for authors in
practice).
> Finally, as a general observation I would point out that BibTeX and
> Refer are *not* the same format,
sorry.. I never used Refer, and I just accepted the word of others.
> and that neither of them is a
> full-fledged "standard". PAPYRUS can import bibliographic lists in
> either of these general styles, but as a rule the PAPYRUS import
> template that works for one user of "Refer" files will not work on
> someone else's "Refer" files without at least some minor changes.
> (This has been true even when both Refer files come from the same
> source, namely EndNote.) BibTeX suffers from the same incompleteness
> of definition.
>
> -- Dave Goldman (RSD@applelink.apple.com)
> Research Software Design
Tell me more about PAPYRUS.
Other tools/formats have been proposed (e.g. TIB).
And have you looked at the UNESCO encyclopedia [:->] on the topic.
P.S. I had written a more structured reply when the computer crashed
on me. I had no time to repeat all of it, and this second version is
not quite as good.
