On 6/7/12 3:12 AM, Heidrun Wiesenmüller wrote:
Correct me if I've misunderstood how Anglo-American catalogs work.
I've just tried it out in your own catalog: Typing in, e.g. "lew
tolstoi war peace" in "keyword" doesn't give me even one edition, let
alone all of them ... The only way to reach my goal via the name-title
string seems to be to use the browse index with the option "author"
and type in "tolstoi lew" (not: "lew tolstoi", of course). Then I'll
have to click on the link "SEE: Tolstoy, Leo, graf, 1828-1910". This
brings me to a new section of the browse menu. Now I have to click
four more times on "next 10 headings", until finally I can see an
entry "Tolstoy, Leo, graf, 1828-1910. War and peace", under which
there is a link "SEE: Tolstoy, Leo, graf, 1828-1910. Voina i mir.
English". Clicking on this again brings me to a new section of the
browse menu with the entry "Tolstoy, Leo, graf, 1828-1910. Voina i
mir. English". One final click on this and - voila! - there are 16 hits.
I think this is more than how "A-A catalogs" work, but is inherent in
the authority data that is being created. In the early 80's when I was
working on the design for the Univ. of California system we tried to
integrate the authority records with the bibliographic records. We
discovered that the cross references were only on the name authority
record, not the name/title record, or, in the case of corporate names,
the record for the first portion of the name. (e.g. "U.S. Dept. of
State" did not have a cross reference from "United States. Department of
State", only the entry for "U.S." alone was linked to "United States".)
We considered doing string matches and proliferating the cross
references over the pre-composed headings, but realized that updates to
those records coming from LC would over-write our changes, and we didn't
at the time have the computing power to re-check all such potential
links every time we got an authority update.
The fact that the production of name authorities (and subject
authorities as well, which followed a similar pattern) did not change
from their card catalog format made it very hard to integrate these into
systems. This may be an area where the wide-spread sharing of data was
something of a hindrance, since none of us could change unless LC (and
NACO, the name authority consortium) changed.
I don't know what your authority records look like nor how they are
exchanged and updated. Perhaps that's another difference, and something
we could learn from German libraries?
kc
This may be o.k. for librarians and some specialist users - but it is
simply awful to present ordinary people with such a difficult and
cumbersome way of reaching their goal. The basic problem behind all
this is, I believe, the Anglo-American technique of using text strings
for collocation, and the fact, that there is no real path between the
different records. E.g. the variant names for Leo Tolstoy are only in
the authority record for this person, but not in the authority record
for "Tolstoy, Leo, graf, 1828-1910. Voĭna i mir. English" as well. But
still: All the information we need is already there, somewhere. So
it's only a question of better data processing.
It is technically possible to transform authorized text strings into
real links. As I said in an earlier mail, we just did that in Germany
for broader and related terms in our subject authority headings. But,
as I've also already pointed out, you don't necessarily have to
introduce record linking. An alternative possibility would be to
expand the title records with the relevant information taken from the
authority records, and then use this e.g. for indexing in a catalog
interface based on search engine technology (like VuFind). In this
case, all the text strings could be kept as they are in the underlying
MARC data. They would be used as "pointers" for the expansion process.
E.g. when there is an authorized string "Tolstoy, Leo, graf,
1828-1910", this tells the algorithm to expand it with all the variant
names from the authority record ("Tolstoi, Lew", "Tolstoj, Lav
Nikolajević" a.s.o.).
I'm sorry if I've got carried away a bit... But I've so often seen the
look of shock and disbelief in the eyes of my students when I
demonstrate a keyword search for e.g. the odes of Horace in the LC
catalog, using a variant name like the Latin "Horatius". This
inevitably leads to zero hits and the funny notice "Please note: The
Library of Congress does not keep a copy of every title ever
published." The students, of course, are used of getting the same
number of hits, regardless of whether they use the preferred or a
variant name in a keyword search, as it is standard in German catalogs.
As somebody who hasn't grown up with the Anglo-American way of
cataloguing, when I look at it from the outside I am both awed and
dismayed. I am awed because you take so much more effort, and include
so much more information in your cataloguing data than we do. To take
just one example: According to the German cataloguing code only the
first editor gets an access point, whereas it's up to three in AACR2.
So, I am very favourably impressed by the richness of Anglo-American
cataloguing data, and I often wish we had the same attitude towards
cataloguing here.
On the other hand, I am sometimes a bit dismayed when I look at the
way this great data is stored, handled and processed. Compared with
conventions and practices here, to me it often seems rather
inefficient and not really suited to this day and age. Collocation via
text strings is only one example. Another would be authority
maintenance, where - if I understand correctly - our customs are quite
different: We're used to automatic updating processes. Whenever a
heading is changed or a variant name added, this only has to be done
once, in the national authority file. It will then be automatically
copied to the correspondent authority files in the regional networks,
and from there, the changed data will be delivered to the local
library systems, again automatically. Due to this system and the links
between authority records and title records, there is no need for any
locally done cleanup. And then, of course, there is the MARC format.
When I teach MARC to students, who are already familiar with another
input format (the PICA format used in the Southwestern German Library
Network), they find it very hard to understand why suddenly they have
to input ISBD punctuation and must type in a parallel title twice (in
245 $b and 246). "Isn't this superfluous?", they ask. "Why doesn't the
machine do it for me?".
Heidrun
--
Karen Coyle
kco...@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet