CATS at Newcastle University
| Published date | 01 March 1988 |
| DOI | https://doi.org/10.1108/eb040396 |
| Date | 01 March 1988 |
| Pages | 26-27 |
| Author | Stephanie Barber,Elizabeth Harbord |
| Subject Matter | Information & knowledge management |
26 -
VINE 72
CATS at Newcastle University
by Stephanie Barber and Elizabeth
Harbord,
Newcastle University Library
N
ewcastle University Library has been
using OCLC's LS/2000 for cataloguing
and an OPAC since March 1985. To provide
greater network access to the OPAC, it was
decided
in 1987 to use the CATS
software (from
Cambridge University Library) on the Univer-
sity mainframe (an Amdahl running MTS) to
provide an alternative
OPAC,
networked round
the campus and over
JANET.
This OPAC
is part
of the University Information Service. The
CATS search program
is
written in FORTRAN
77 and provides public access searching of the
catalogue by keywords taken from author, title
and subject headings in
the MARC
record.
(See
VINE 63 and 47 for further details.) We have
thereby gained the opportunity to compare two
different systems running
on the same
database,
and also a back-up OPAC if LS/2000 is not
available, for example during maintenance.
Our database consisted at the time of loading
into the mainframe of approximately 210,000
records
in
LC
MARC
format (including Library
of Congress subject headings or MeSH) for
books, serials and audio-visual materials. The
records were copied to eight tapes for loading
on the University mainframe. Records are held
in
CATS
format which replaces
the MARC
tags
with a single letter code to save disk storage
space. There are size limits for the MARC
record before and after translation into CATS
format. After the translation from LC MARC
format
to CATS
format the records
were
loaded
sequentially
-
i.e.
they were written into position
in the machine storage and the indexes were
written to separate files. When all the records
had been added
the
index files
were
merged and
sorted. This was a large computer sort task, but
we felt that the cpu times involved in loading
records
in
'indexed'
mode,
which involves find-
ing the correct place in the index to insert each
entry, were prohibitive. The MTS sorting
routines are very good and there was in fact no
difficulty in sorting and merging more than 1
million index entries.
The records we loaded initially generated near-
ly 4 million index
terms.
The index terms were
distributed as follows:-
title words-1,780,015
author words-
1,191,572
subject words - 691,349
plus the index of 210,000 item numbers.
We loaded the original database at the rate of
c.20,000
records per
job.
There was a trade-off
between
the
amount of
cpu time
a job could take
without producing an unreasonable elapse time
(on
an
overnight run the longer
the
cpu time re-
quested by the job the later it would be
scheduled to run!) and the need to keep down
the number of index files generated for later
sorting, since each run generated its own set of
four indexes.
Updates have
to be done in
'indexed' mode now
that the database records and sorted
indexes
are
in
place.
New records
are
added to
the
end of the
database file, but index terms
are
inserted
in
the
corret place in the index files. The database is
updated monthly by tape from LS/2000 (c.l-
2,000
records).
The main changes made in our version of
CATS,
which is not the same as that currently in use at
Cambridge, are as follows:-
1.
I/O reads and writes to allow for the MTS
operating system way of file
handling.
This
was done by John Bagnail (formerly of
Newcastle University Library) and Sam
To continue reading
Request your trial