Authority Control in the 21st Century: An Invitational Conference


Practicalities of Online Authority Maintenance: Based on the University of Tennessee, Knoxville Libraries Experience

Agnes M. Grady

University of Tennessee


Table of Contents


I. Introduction and Background

The beginning of my cataloging career at Oregon State University coincided with the implementation of AACR2 and the resulting massive amount of changes to name forms. Thus when in 1984 the Authority Control In an Online Environment Interest Group (ACIG) was formed by two ALA divisions, I knew this would be one of the most beneficial catalog maintenance organizations to be involved with. Consequently I served as chair of ACIG's subgroup on online maintenance for a few years. My interest in authority maintenance has continued throughout my career, and I am currently working to implement effective and efficient authority work in a local system online environment.

Few libraries could afford to provide complete authority maintenance work in the card catalog environment. Most libraries did expect that their original catalogers would create name forms according to the cataloging rules. Many catalogers also generated the reference structure that helped the catalog users to find the appropriate heading. Series authority work was generally held to be essential since a great deal of money was expended on acquiring serials and efficient retrieval of these materials required consistent treatment. Maintenance of subject access was usually the place where short cuts or minimal work was undertaken.

The online environment offered hope for those catalog librarians who felt that authority work in all aspects (names, series, and subjects) should and can be kept on a current basis. A computer can provide reports about headings that can be seen as a filtering device: headings in bibliographic records both matched and unmatched by authority records and headings for which the computer makes changes based on authority records. These reports can be reviewed by cataloging staff at appropriate levels another filtering mechanism so that personnel resources are efficiently expended.

I have been working for several years in applying authority control over a bibliographic database that was created by various methods attributable to the nature of the local system in use at the University of Tennessee, Knoxville, Libraries and to budgetary decisions. A wide range of library staff have made use of OCLC as the major source of our bibliographic records. Professional catalogers and staff have had a hand in both retrospective and current cataloging. Student workers had a role in a severalyear reconversion project. In the early 1980s brief record keying in the local system resulted in acquiring REMARC records from Carrollton Press (these records had a dual purpose, the second for allowing books to circulate). As a result, the existing database was generated through means of machine and by staff having various levels of cataloging expertise, particularly related to authorized headings. Except for the Library of Congress records acquired in recent years, the database cannot be deemed to have consistency in all of its headings. It is this lack of consistency in the earlier bibliographic records that has driven the focus of our authority efforts while the practicalities of automated versus manual authority processes has shaped the methodology.

UTK Libraries currently has the Geac Library Information System (GLIS) 9000 for cataloging but began with GLIS 8000. The move from the 8000 to the 9000 was to include authority features within the cataloging module. In fact, UTK Libraries was a beta test site for Geac's linkingloader program written for GLIS 9000. Many of the library decisions about applying authority control to the existing database made in the late 1980s were dependant upon information we were provided during the development of this program. The Libraries bibliographic database was sent to an authority vendor in late October 1991. After processing by the vendor, we received a small percentage of bibliographic records to be reloaded and about 400,000 authority records that matched headings in the bibliographic database. To make it plain, the bibliographic database was not upgraded to current heading standards by reloading corrected bibliographic records. Instead that upgrading would be done by the Libraries running Geac's linkingloader and by Cataloging staff reviewing the results of that work, identifying and making links, corrections, and other necessary editing on those headings that the machine could not or did not link or change.

The authority records from the vendor were grouped in categories based on type of record and the presence of 4xx tags, a grouping we hoped would facilitate the linking process. Except for series and split heading authority records, each contained tags numbered 999 that had information about each heading in the bibliographic database that matched either the authorized heading or a see reference. This information (the bibliographic record number that contained the heading, the tag number of that heading and the occurrence number of that tag) was to be used by the linkingloader to quickly replace the text of that heading in the bibliographic record with a link to the authority record, thus authorizing that heading.

After reloading the relatively small number of changed bibliographic records, the Libraries began machine authority processing with subject authority records. Only those bibliographic records having the exact authority heading were changed, thereby leaving most subject headings with subdivisions unchanged. Cataloging staff began their authority work by reviewing the reports generated by system processing of the authority records. These reports allowed staff to find those subject headings that were not processed because the bibliographic heading was not correctly tagged, contents were misspelled, and those headings with subdivisions that were left dangling under the obsolete heading.

Although our initial decision was to load subject authority records, process them, then load name authority records with references and process them, etc. etc., I realized soon after the processing of subject ARs began that Cataloging staff needed to know what was in our entire authority database. So after the processing of subject authority records was finished, all authority records were loaded before machine processing was begun again. Name/uniform title authority processing took place over a period of many, many months. [Chronology: subject loading = 6 months, name loading = 4 months (the result of improvements made by Geac to the linkingloader as well as scheduling and other adjustments made by Library Systems to linkingloader to speed up the loading) subject processing = 10 months, name processing = 18 months]

II. Principles of Effort with Personnel and Machine Resources

With the advent of online authority work, everyone in Cataloging participated to some extent. Original catalogers dealt with split headings and were consulted about validity and appropriateness of headings by the support staff. Adaptive catalogers, the highest level of support staff, reviewed the systemgenerated reports, resolved machineidentified problems, completed linking manually, and corrected tag numbering and tag content. Library specialists were given voluminous headings needing manual tag number and content corrections that were identified by both original and adaptive catalogers. Training in heading assignment was essential to allow this division of labor without getting the original catalogers overwhelmed by questions.

For subject authority, I undertook the review of headings having the term "Negroes." This work was not only a split heading situation but also included subdivisions having that term. After name authority records were loaded and processed by the linkingloader and because of the complexities, I decided to work on all United States entries: names, subjects, and series. It is from this experience that I've determined the following: since terminology is constantly changing in society, subject headings must keep up to be relevant and corporate body names will always need attention.

Pseudo Authority Records and Pseudo Cross References

I soon realized that when staff identified obsolete headings with subdivisions, changing all records individually would take too much time but that the system processing could get bogged down if we stopped the initial processing of authority records. A compromise was reached by setting an upper limit on the number of bibliographic records containing the same authority conflict that a person would manually change. When this limit was reached, action was taken to incorporate machine processing. This action might be creating the needed authority record online in an abbreviated form with just the established heading and any needed crossreference headings that would replace obsolete information. Or, in the case where we did have an authority record but had bibliographic headings with an obsolete form not included as a 4xx reference, we would add a reference tag and reprocess the authority record. These pseudo authority records and references were the only way to have a search and replace technique. The normal workflow in Systems would be interrupted periodically to process (or reprocess) these individual records. [Needless to say, there was close communication on everything to do with authority processing between Cataloging and Library Systems!] Unfortunately, the mistagged headings could not be handled by the machine. Staff had to renumber the tag and as long as the record was being edited, completed the linking manually.

III. Authority Record Content: Its Effect within an OPAC from the Library User's Point of View

Retrieval is the name of the game and is the overarching reason for devoting time and effort to authority work as far as I am concerned. Providing intellectural access to a library's collection is the purpose of a catalog and authority work is essential for collocation and giving the catalog users clues to headings that retrieve citation information that fill the users needs.

Cluttered Index Screens

Authority records retain a great deal of the past. The most unfortunate aspect of cataloging history carried in authority records are the references of pre-AACR2 heading forms that existed in the card catalog environment. I always think of "arcane" when it comes to filing in the card catalog: punctuation defined types of headings that most noncataloging librarians failed to understand, let along the public. The literal computer put paid to all of that! So if your local system is not sophisticated in the authority record realm (and most are not), this means that a heading distinguished only by a bit of punctuation from its reference will file in the computer right next to it.

The GLIS 9000 does allow references to be hidden from the OPAC. Part of the authority work done by Cataloging staff was to hid those see references that not only cluttered the index screens but actually hindered the catalog user from quickly finding the entry being sought.

IV. Authority Work as a Special Reconversion Project

I view authority work on an existing database (as opposed to authorizing records being added to the database for current cataloging) as a retrospective conversion project. As such, you are dealing with information without the piece in hand. Thus you may find that a subject heading that just fits was not assigned because it didn't exist at the timea title was first cataloged. The original cataloger who workied on the split heading Crime and criminals can attest to this. Sometimes the added title entry is a clue to the need for this subject heading. You may also find an obsolete heading no longer in any authority file and wonder what to do. When we began our authority work, "Revised Library of Congress Subject Headings" helped identify subject headings made obsolete from 19761990 and gave their current forms. In 1994 we were pleased to see Joyce Ogden's compilation entitlted "Library of Congress Changed Subject Heading Subdivisions." The cataloging tools used in current heading assignment such as CDMARC Subjects, LCSH, FreeFloating Subdivisions, LC Period Subdivisions under Names of Places, and Subject Cataloging Manual: Subject Headings have all gone through several editions since our project began.

Systematic Review

After completing all reports generated by system authority processing, adaptive catalogers were given a letter of the alphabet to review the headings index in a systematic way. When working only on subject authority, they got halfway through the alphabet before name authority processing began. At that time, we turned our concentration to names because reference librarians believed discrepancies resulting from linked and unlinked names needed high priority resolution. Since the alphabet did not make much sense for addressing a priority for working on names and because we had recently finished a fiction reclassification project, we began reviewing the names of literary authors, philosophers, politicians, and composers. Once we felt that most wellknown authors had been reviewed, we went back to the alphabet. With names we had two indexes to review: names as names and names as subjects.

Touch a Record for any Editing, then Look at Everything in It

From the beginning of authority maintenance work, the principle of looking at an entire record once it had to be manually edited for any reason was adhered to. Although I could say that limitations of system resources in processing edited records was the sole consideration behind this decision, it is not true. The main reason was my strongly held conviction that repeatedly returning to a record, on a tag by tag basis, was a monumental waste of human resources. However, the expertise of the staff and the training provided were essential quality factors to this principle since "looking at a record" does not necessarily equate to the result of the record having all valid current headings.

Series

The issue of series information in bibliographic records and linking to appropriate authority records appeared early in UTK Libraries investigation of an authority vendor. Headings for series have a unique complexity that need careful attention. PreAACR2 series forms not only use pronouns in description, but MARC tag numbers are different with most AACR2 authorized headings. Since most Cataloging support staff were learning as they were going about headings in greater depth than they ever knew before, I decided to leave series until after subjects and names were completed. From my experience with series headings encountered in my review of "United States..." name headings, I could see where guidance in this area was needed. At this point in time, we have not begun series authority work in an ongoing basis.

V. Challenges Continuing, Caveats Contributed, and Conclusions Considered

As I reviewed my file cabinet full of folders containing all the paper documentation I have kept on this authority project for the last decade and tried to distill all that activity for this paper, I tried to focus on what generalizations would naturally enfold.

Keeping Existing Authority Records Current

It was not long after Cataloging began its online authority work with subjects that I began investigating the potential changes to our authority database. I had the revised and cancelled subject headings listed in Cataloging Service Bulletin. #59 (issued Winter 1993) checked against our authority database. Those changes were significant enough to begin editing our existing authority records according to information provided in the Cataloging Service Bulletin. However, we soon had to cease this activity. Although the mechanism for the global updating of headings had functioned in our early testing of the linkingloader, this process did not always function properly, particularly when a large number of linked bibliographic records were involved. Geac had to investigate several, similar problems related to the authority features that arose during the course of reallife operations. These problems and the pressure to complete the batch processing of authority records did not allow us to proceed fully with what I would consider normal authority record maintenance work.

Getting New Authority Records

My desire to get to the place where UTK Libraries was getting new authority records based on headings in new cataloging had to take second place to my responsibilities and involvement with librarywide activities for getting a second generation automation system. That process put on hold my investigation of the possibilities. But easy access to new and revised authority records for authority processing is essential for several reasons. Those of us who have been responsible for maintaining catalogs have always known the dynamic nature of our catalogs, no matter how large or small. We have had experience of coping with constant change. Authority records are one primary factor of this change. Existing authority records are subject to change and new authority records are being created daily. When you see a valid heading in a currently received bibliographic record changed to an obsolete heading by one of your existing outofdate authority records, the impetus to move on your static authority database becomes imperative.

If I Had It to Do All Over, or, You Can Be So Smart with Hind Sight!

The lengthy process and activities I have described has been challenging to say the least. And the authority work on UTK Libraries' bibliographic database has been only part of the picture of database maintenance for the sake of our users. Projects in addition to authority work include reclassification of works of fiction (about 25,000 titles) and Reference titles classed in Z over 1200, providing summary holding statements for serial titles, and upgrading brief records that were keyed for being overlayed by records from Carrolton Press but for which no record matched (about 35,000). In all of these projects, we were able to get rid of several thousand no longer needed bibliographic records since we were taking inventory of these collection pockets at the same time. Because these projects help ensure as accurate a database for our holdings as possible, our migration to another system should be facilitated.

My presentation may reflect undue influence from problems with our local system. You may think my experience contains elements that others may not experience, those related to a particular local system. But in my considered judgment, all of us must contentwith how a local system incorporates authority processing. The details may vary, but there will be problems. Local system vendors must become actively involved in venues like ACIG in order to comprehand what is needed to allow Cataloging staff to be as productive as possible and let the machine do the routine. And this must be done without subverting an efficient and effective workflow. Vendor user groups may be a key to keeping vendors aware of changes in cataloging codes and other standards that may impact authority work.

The type of authority work I have described is very laborintensive and redundant in that it requires human intervention on a microlevel, including the calling up and examination of every heading in bibliographic records that have already been handled at the time of retrospective conversion. Although this type of authority work is appropriate and integral to a Cataloging unit's mission, retrospective authority work is one activity that is on the top of my list for outsourcing. Furthermore, let the machine do what it can do, but allow humans to do what they do best. The University of Tennessee, Knoxville, Libraries experience indicates that there will always be plenty of database maintenance left for cataloging people with their intelligence and intuition that no machine can match. It is my hope that, nationally, this will be the direction that will be taken in regards to authority work.


Appendix 1: University of Tennessee, Knoxville Libraries Authority Processing Chronology

March 1986 - Spring 1988: UTK Libraries' investigation of authority processing by commercial sources resulted in written specifications that were distributed for bidding by vendors. Contract signed. Test on 5,000 bibliographic record database was completed by vendor and evaluated by UTK Libraries Cataloging.

Spring 1987 - Spring 1991: Software program for linking loader for GLIS 9000 developed by Geac and betatested by UTK Libraries in Spring 1991.

Spring 1991 - October 1991: UTK Libraries revised specifications for authority processing based on beta work and authority vendor's initial work. Major changes focused on series authority processing and reducing the work of bibliographic record replacement to absolute essentials and upgrading that could not be done from information in authority records.

October 1991 - February 1992: UTK Libraries sent bibliographic records to authority vendor; vendor returned authority records that matched bibliographic headings and 41,820 (4.6% of entire database) bibliographic records that were changed according to specifications.

Number of Bibliographic Records Sent to Vendor:895,069
Number of Authority Records Returned by Vendor
Name:348,260
Subject:56,784
Uniform Titles (Series):14,273

March 1992 - April 1993: UTK Libraries loaded reprocessed bibliographic records and subject authority records. Using the linking loader, UTK Libraries processed subject authority records. Cataloging staff become immersed in online authority work.

April 1993 - August 1993: UTK Libraries loaded name and uniform title authority records.

August 1993 - February 1995: UTK Libraries processed name and uniform title authority records via linking loader. Cataloging staff turned their authority attention to names.


Appendix 2: Statistics of Cataloging's Authority Work

Categories1992/931993/941994/951995/96Total
Authority Records
Enhanced and Changed (1)3,6941,86895355,692
Created (2)10866466226
Deleted59636101
4xx Tags Hidden2,1417436,4232,96312,290
999 Tags Deleted8512,7631,9288616,403
Bibliographic Records
Content Corrected14,58823,42313,8803,17955,065
Tags Renumbered (e.g., 650 to 651)2,6302,8772,8469029,255
New Tags Added26937117329752
Links Between Authority Records
and Bibliographic Headings
Completed6,9137,34010,5822,82027,655
Cut6947913154651,950
Other Systematic Reviews Undertaken
Subject Headings Splits (3)137
System-generated Reports5,6301,7287,358
Personal Author Names4,396584,454

Notes:

  1. Linking-loader detected errors account for the large number 1992-94
  2. Brief records (aka "pseudo" authority records)
  3. "Negro" headings & subdivisions under names not included in this count

Year designation is for Fiscal Year beginning with July and ending with June


Return to Proceedings Home Page


Advanced Search | Careers at OCLC | Feedback | Privacy Policy | ISO 9001 Certificate | ©2003 OCLC Online Computer Library Center, Inc.