83VINE

VINE

Vine Volume 13 Issue 3 1983

Page 13 -18

RECON AND REMARC AT EDINBURGH UNIVERSITY LIBRARY

Abstract: Introduces the cataloguing, catalogue enquiry and circulation control system based on a Geac 8000 mini being installed at Edinburgh University Library. Cataloguing presents the most urgent problems since all current cataloguing is manual and decentralised, with no union catalogue or machine readable file. The retrospective conversion and database creation programme is to take 5 years, with data being bought in from external sources wherever possible, for economic reasons, since the internally maintained catalogues lack any degree of uniformity or consistency. The initial phrase of database creation involves the use of the American database REMARC from where Edinburgh has been obtaining external records which can be supplied in UKMARC format. Requests are keyed directly from their shelf lists using 29-character search keys. Edinburgh aims to key some 400,000 requests annually and expects a hit rate of around 50%. The major benefits anticipated from the REMARC system are a greater consistency of headings and cataloguing and the fact that the provision of LC subject headings on the REMARC records will allow comprehensive subject access over the complete file. The second phase is an online check on the SCOLCAP/BLAISE database and further phases are currently under investigation. Online cataloguing is scheduled to begin over the summer and the circulation system is due to go live over Easter 1984.

Introduction

Currently celebrating its 400th anniversary, Edinburgh University is one of the oldest academic institutions in the country. As far as the library and its services are concerned, the years have seen the evolution of a fairly complicated organisation, with numerous branches and service points scattered across the city-wide campus. Any automation project is, therefore, likely to present numerous difficulties both in scale and in reconciling current practices. Add to this the sophisticated computing environment of the university which has promoted awareness of what can be achieved using new technology and we have the basis of what promises in the long-term to be a highly ambitious and exciting, if rather complex, project. The library is installing a Geac-based cataloguing, catalogue-enquiry and circulation control system. In addition, the Geac 8000 machine will be a host on the Edinburgh Regional Computing Centre (ERCC) Network, thus providing network-wide access to the library's files. Since the ERCC already encompasses the universities of Glasgow and Strathclyde (the first of which is also installing a Geac Library System) the potential user community is considerably larger than that being served directly at present. Automation is seen not just as a means of making manual methods more efficient but as a positive opportunity to advance and extend the services offered by the library.

Configuration

The hardware supplied will consist of a Geac 8000 minicomputer with 10 x 150 Mb disks to accommodate records for the 1.8 million items and allow the

Page 14

necessary overheads needed for automatic running. There will be 60 enquiry terminals (11 for staff use) of which 20 will have printers attached for on-demand hard-copy output. Some of the public terminals will additionally be fitted with light-pens to allow both the usual Geac reservations and borrower enquiry facilities and self-service issues and returns - this is to continue what is already current practice in some of the low-staffed departmental libraries. The ERCC network can presently be accessed from more than 1100 terminals.

Current practice

To date, Edinburgh has not automated any of its main housekeeping procedures, apart from using the Network's Viewdata facilities to mount files of new accessions and of periodical holdings (see VINE 45). Loans can be made from any of 27 branches, each of which has its own loan policies. Although with the introduction of Geac it is hoped rationalise some of the more extreme differences, the automated system will still be required to cope with around 20 different options. Effectively, for circulation, the system will be a shared one, not unlike that currently installed at the University of London.

Cataloguing present further and more immediate problems: all cataloguing is manual and decentralised; there is no union catalogue; and, with the exception of a file of ISBNs derived from the accessions listing, no machine-readable file. Creating the database is, therefore, a task of no small order. The retrospective conversion programme will run for the next 5 years, using, hopefully, staff funded under the MSC scheme; it will be headed by a project team of senior library staff, on a two-years' secondment from their normal library duties, by the UGC with a re-structuring grant. Over the 5 years, VINE will report on progress generally and on specific aspects of the system as they are installed and go live. In this article, now that the broad lines of the project have been laid down, I propose to focus on the retrospective conversion programme which got under way in February.

RECON

The basic premise behind Edinburgh's recon strategy is that wherever possible data should be bought in from external sources, with local EMMA cataloguing seen very definitely as the option of last resort. The reason for this is economic. Edinburgh's file will contain data taken from a number of independently maintained catalogues which, though internally consistent, lack any overall uniformity. Edinburgh's Geac database is not, however, a second bibliographic file of short circulation-control records but the authoritative catalogue requiring the imposition of some consistency and uniformity of entry and heading on to the existing records. If existing entries were to be keyboarded directly on to the database, a considerable amount of editing, of updating and amending headings, and of authority checking would be needed from staff with professional library skills. These are in comparatively short supply and are expensive. However, there is a pool of unskilled (in professional library terms) labour which with the sponsorship of bodies like the MSC can be used to keyboard requests for externally-supplied records.

Page 15

Edinburgh, therefore, intends to follow the 4-point plan, detailed below; it will be a lengthy and time-consuming procedure with each stage involving a new pass of the catalogue but is expected to be cheaper than currently available alternative methods.

1. Use of REMARC and LC MARC records

2. Check for SCOLCAP records online

3. Go to other external sources such as OCLC

4. Input as EMMA online to the Geac database, using the Geac cataloguing module.

REMARC: GENERAL OUTLINE

Since April Edinburgh has been submitting requests for records from the REMARC database, the first library outside the US to do so. REMARC is a service offered by Carrollton Press Inc. - available in the UK through Chadwyck-Healey Limited. It began as a project to produce a Title Index to the Library of Congress shelf-list; the potential of the data they keyboarded for this was rapidly realised and Carrollton has nearly completed the input of some 5 million non-MARC records. Records were originally entered in alphabetical order by title but in mid-1982 the order of input was changed to concentrate on English-language records; this portion of the database was duly completed in April 1983, and the foreign language section is scheduled for completion early in 1984. As far as cataloguing content and standards are concerned, Carrollton Press uses the shelf-list card as the authority. The cataloguing at the time that the card was created is kept in its entirety; the choice and form of entry, the punctuation in the body of the card and the subject headings established are all keyed in directly as they appear. Although LC began to create MARC records for English language items from 1968 onwards, French did not follow until 1974, German, Spanish and Portuguese until 1976, and Dutch, Scandinavian, Italian and Romanian in 1977; the result is that since 50% of LC holdings are non-English language over 20% of REMARC records are post-1970.

The initial key-boarding - from facsimiles of the shelf-list - is coincidentally also carried out in Scotland, in Irvine. The data is keyed to disk and then corrected and edited before being sent on tape to LC for passing through its format recognition programs for allocation of MARC tags. Resultant tagging errors are corrected by Carrollton staff in California. As full a MARC record as catalogue content allows is what is aimed for; the only fields listed as being completely excluded are: 082 (Dewey No: very often not present on the original); 490 (untraced series title); 5xx (notes); 020 (ISBN); 086 (Supt. of Docs. number); and 350 (price). Of those, only the exclusion of the Dewey number affects retrieval. Full details and a fuller technical description can be had from Chadwyck- Healey Limited. However, for UK libraries, REMARC records can be converted and supplied in UK MARC format.

Page 16

Though the REMARC files are available online via Dialog for anyone interested in viewing them, the basic service is batch and offline. Libraries submit requests on floppy disks which are read to tape by Carrollton and then run against the REMARC files. The customer-library is then supplied quarterly with a tape of hits plus a series of reports listing:

(1) errors detected in the keyed algorithms

(2) listing of multiple-records responses to title key requests - these are further investigated by the library and the request is re-submitted once the correct record has been identified.

Costs are incurred only for records received i.e. for 'hits'. Currently charges are £0.34 for each REMARC record, with a further optional charge of £0.03 for each record converted to UK MARC format; and £0.12 for each LC MARC record (the LC MARC files are also mounted and can be searched concurrently).

No other costs are imposed: the Apple micros and floppy disks used to keyboard requests are all supplied by Chadwyck-Healey, who provide as many machines as can be kept occupied on a full-time basis. The operation has been kept as simple as possible and normally an operator would expect to key around 100 requests an hour in response to a simple sequence of prompts from the Apple. Search keys are of two types: either the LCCN can be input plus a checking code; or, alternatively, a 29-character algorithm is used, comprising the first 23 characters of the title (including any articles, but excluding English articles) plus three optional characters from the place, and three from the date, if known. The software both prompts the operator and "normalises" the input to the exact format required. Finally, the library can add any local data (e.g. Dewey no).

Use of REMARC at Edinburgh

Input is being keyed by MSC staff who number 39; this work force is divided into 5 input teams plus an administrative, editorial group. Requests are keyed directly from Edinburgh's shelf-list, using, in the majority of cases, the longer, 29-character search key. The accuracy and efficiency of this will be subjected to its first real test at Edinburgh, since the US libraries using REMARC have generally used the LCCN as search-key. In a few instances, and at the discretion of the MSC staff, the National Union Catalogue is consulted to find the LCCN - this is needed mainly for journals and occasional common titles where the first 23 characters of the title are non-significant e.g. Cambridge History of etc. In the first two months, over 100,000 requests were keyed though this rate will drop as more time has to be given to editing, to checking error and multiple hit reports, and to checking Edinburgh's own database for duplicates - this last step will be introduced once the first library's catalogue has been processed and mounted on the Geac. However, on average Edinburgh is working towards an annual rate of 400,000 records.

The first large batch of keys has now been checked and some interesting pointers are beginning to emerge. Trial runs of a few hundred records had

Page 17

suggested a hit rate of 50%, though for some areas of stock it is impossible to predict. The first batch of 31,000 records produced almost 18,000 hits; a hit rate of 58%. Almost 1000 of these were multiple hits. These fall into two categories. If the hits per request number 10 or less, then bibliographic details are printed out on the report, making identification of the correct record comparatively easy, particularly as the majority of cases - 796 – had only 2 or 3 hits per request. Only 30 keys retrieved 11 or more records, and had no bibliographic details reported, entailing further work on Edinburgh's part. A further 741 records were reported to have errors of various types and were not matched. Most of these can be resubmitted, increasing the final hit rate by a further 1.5-2%. The first batch consisted of Science, medical and veterinary collections and part of the undergraduate collection. This latter group no doubt helped push the hit rate beyond the estimated 50%, but is none the less encouraging.

The records are supplied in shelf-mark order and the conversion teams then have to go through the catalogue again to identify non-hits and to report any glaring errors. This task is less of a nuisance than it might be since, at the same time, the Geac item number, held in bar-code on the shelf list, is wanded to create the necessary link between bibliographic and item record on the database. Edinburgh expects two major benefits from adopting the REMARC approach. Firstly, the consistency of headings and of cataloguing will be much greater than could possibly be achieved in-house; and secondly, since Edinburgh has 9 different classification schemes, which will all be retained, the provision of LC subject headings on the REMARC records will allow, for the first time, comprehensive subject access over the whole file and over all the library's stock.

Future RECON steps

Once the REMARC records have been received, Edinburgh will enter the second phase which is to do an online check of the SCOLCAP/BLAISE database. If a record is found, then it will be flagged for dispatch to Edinburgh via a weekly tape for loading on the Geac. Effectively, this is the system which Edinburgh proposes to use for some at least of its current cataloguing viz. for recent UK material and for items which it might expect to find on SCOLCAP because of overlap of interest with other members. Step 3 - the possibility of using other databases, particularly OCLC via its new Selective Record Service (see page 19) - is still under investigation though it is regarded as a strong possibility.

Networking

The proposal is that Geac is to develop the protocol to interface the Geac to the ERCC network so that network users will be able to go straight into catalogue enquiry on the Geac. Other options are also being considered, however. These both involve mounting copies of the catalogue files on other machines, either on one of the ERCC's ICL 2900s or on a MEMEX currently

Page 18

undergoing evaluation. The main benefit of such duplication would be to divert catalogue enquiries and to protect the Geac system, particularly the circulation control functions, from degradation of response time if catalogue enquiry use becomes unexpectedly heavy. There are further dimensions to the way Edinburgh will use networks. The borrower records will be based on data provided from Administration's files; change of address, however, is normally notified first (and often solely) to the library which will, in future, easily transfer this information back to Administration. Links will also be established with SCOLCAP and with SCOLCAP member libraries. Dial-up access to Edinburgh's database will be provided so that libraries can search for and then request copies of records held there; these will be transferred to the SCOLCAP database, passing on a second-time use charge of not less than £0.13 if a REMARC record is involved.

Conclusions and Timetable

A large number of very different ingredients is going into the Edinburgh system and much of the future interest will lie not only in seeing how well each part of the system works but in assessing how effectively the various aspects and services combine. At the time of writing, the Geac 8000 had been delivered with the cataloguing software, ready to undergo acceptance trials; file transfer protocol software is scheduled for early August, and this will allow the network link. Online cataloguing for current materials - local data entry and tape-loaded from SCOLCAP - will begin over the summer. The circulation system is due to go live over Easter 1984 with the undergraduate collection, followed by the Science, Medical and Veterinary Libraries over the summer.

Information supplied by: Mr. Derek Law, Library Automation Controller