Authority Control in the 21st Century: An Invitational Conference
Authority Control in the 21st Century: An Invitational Conference
My topic this morning of authority control and personal name headings was chosen for the purpose of suggesting an alternative to the current cataloging system which sets as the goal the establishment and control of all personal name headings. This goal, however, of complete authority control has never been completely realized in the typical library catalog. While shortcuts have most often been viewed as temporary measures, their recurring nature contradicts this perspective and pushes us to confront the gap between the total authority control we are trying to achieve and the reality of partial authority control over personal name headings in library catalogs.
In a paper presented at the OCLC forum titled "The Future is Now: Reconciling Change and Continuity in Authority Control," I identified several reasons why it is both important and possible to examine the conceptual foundation of retrieval and therefore of authority control. These reasons include the usefulness of technology as a catalyst for change, our continuing responsibility to make the best use possible of available resources and the cooperative nature of library cataloging operations (Younger 1995b). I had previously suggested that the concept "utility" together with a measure of "useful recall" be introduced to complement "comprehensiveness" and "comprehensive recall" (Younger 1995a). Utility calls for focussing our attention on those personal names for which authority control is most likely to prove significant in effective information retrieval, therefore supporting authority control while suggesting practical limits can be defined and applied. It would provide us with a greater ability to focus authority control work where it is most necessary, thus helping to ensure expenditures that are driven as much by the principles of effective retrieval as they are by the available resources.
Today I am going to talk about how authority control could be done differently, which some have labelled this as "just-in-time" authority work, or "just enough" authority work. Let me briefly talk about the current system as the basis for change.
Our current system of cataloging and authority control functions in a decentralized environment. Cataloging is done in many libraries by many individuals, each one creating bibliographic and authority records. The bibliographic records are sent to one or more bibliographic databases, such as OCLC, RLIN, WLN, and UTLAS, which are respectively both unique and overlapping in their inclusion of bibliographic records. Authority records, when posted for use beyond the local library, are sent to a single master file, developed under the aegis of the Library of Congress. This file is accessible from multiple databases, again including OCLC, RLIN, WLN and UTLAS.
When cataloging, catalogers establish each name heading according to the cataloging rules. In conjunction with applying the rules, the cataloger searches the name, usually in the National Authority File (LC NAF) and in the local catalog, to determine whether the name already exists in the same or variant form, whether it is a new and unique heading, or whether it is a new and nonunique heading. Once the name form is established, the cataloger decides whether to create an authority record taking into account local policies and/or participation in the Name Authority Cooperative Program. If an authority record is created, the cataloger may post it only to a local file, to the national file or both.
This brief description of the current system points to several reasons why total authority control has not been achieved. The first is that new name headings are generally established in the context of the local catalog (when they are not already found in the national authority file) with the result that when used by another library in its copy cataloging operation, the heading may not necessarily be entirely correct in the context of that second local catalog. Thus, even when a national authority record exists, the heading is not necessarily completely controlled in that second catalog. It is true, however, tthe scope of this problem is diminishing as the size of the national authority file increases because the authority file is consulted even before the local catalog, and so if there is an authority record, that form is used (or adjusted as necessary).
Yet, the national authority file remains incomplete because not all cataloging libraries contribute records and even of those libraries who do so, many participants post only authority records for only some of the name headings they establish. The absence of an authority record means there is no automated "control" of that heading in any catalog. And, because we are increasingly dependent on automated authority control, the lack of automated authority control often means no authority control.
The reasons for not creating authority records are important as they can be interpreted as laying groundwork for determining which names do or do not require authority control (at least initially) if we were to consider adoption of a concept of utility. The reasons relate to the heading itself, including situations when no cross references are needed and there is no particular information to record about the heading or to the library's unfamiliarity with constructing name headings and authority records in particular languages or the need to conserve catalogers' time for other work.
If not all personal names required control at the outset of their establishment and use in the catalog, a cataloging system could work this way. Catalogers could create bibliographic records from the document in hand, establish the name according to AACR2 from the title page without referencing the catalog. (Defined exceptions include personal names which are part of the classification, such as authors, artists, philosophers and ahtletes.) The completed bibliographic records would be contributed, as happens now, to one or more bibliographic databases.
Although the cataloger would not be making an authority record, from the bibliographic record, the computer could extract each personal name along with attributes such as language of the text, country and date of publication, and classification number, and create a machine-derived authority record. These authority records might or might not be contributed to the national authority file; their generation at this point in time is to have a record that can be used by the computer in searching and comparing this name to existing name headings found in the name authority file and bibliographic databases. In doing the searching, the computer would probably be able to replicate some of what a cataloger now does. If the name was found to exist in the same form or found not to exist in that form, the computer could accept the name form, update the authority record. If on the other hand, the name was found not to exist in that form, the computer might be asked to print a report of similar names together with their attributes for a cataloger to determine if these were the same name heading or not. I am describing here a system no one yet has. However, computer algorithms do exist for evaluating the degree to which entities and attributes are similar; this is the underlying premise here.
So far this doesn't sound very different from what we do now in using computers to reduce human effort and increase productivity (computers work faster and repetitively without getting bored). Yet there is potential to alter decision making, to reduce the demands on authority work by allowing less than a full exploration of the fit between new and existing names. We could ask the computer to not look at names in particular classifications or types of materials, to consider whether clusters of similar headings occur fewer than five times and in those cases to not pursue further differentiation or amalgamation, that is, to not worry about name headings that appear to match or be similar to an existing heading when the classification is very different, This is not very radical; we are already in a world of partial authority control although that is not the stated goal we wish to achieve. The change is that we would be designing the system toward that end; to leave headings not completely controlled because under specified conditions, we do not believe it will pose a problem to successful retrieval (for a discussion of personal names as handles to information, refer to Younger 1995b).
Beyond an extended role for computers in authority control, the above proposal implies that headings can initially be uncontrolled and then move to a controlled state as necessary to support effective retreival. The conceptual model articulated under the auspices of the Library of Congress Program for Cooperative Cataloging does view authority control as an iterative process and one in which a name heading could be established from the piece in hand and "controlled" in the larger context of the catalog at a later point in time. Dynamic updating of names in authority and bibliographic records, by replacements fields or records, globally rather than singly, is already the preferred mode of operation in the current system; dynamic updating would continue as a fundamental element in authority control.
Authority work makes heavy demands on the cataloging process. I am proposing we examine possiilities for reducing human efforts with greater use of computers and second, that we further reduce our efforts by allowing some personal name headings to enter and exist in the catalog (for some period of time, perhaps forever) without being under full authority control. Despite the fact, however, that uncontrolled personal name headings occur in every catalog, some by default, some by local decisions made categorically, e.g., not to control headings occuring on bibliographic records from ERIC loaded into the local catalog, and some by individually made decisions, any purposeful consideration of partial authority control is unsettling and unsatisfying.
There are several reasons for that, beginning with widely held beliefs that users shouldn't have to do any authority control, they can't do it, and they won't do it. And, because we believe users must always be given the opportunity to choose from all library resources related to the search topic (comprehensive recall), we strive for complete authority control. Library catalogs represent a deliberately assembled collection of related works, related by authors, editions, and/or subject whereas journal articles are related primarily by appearing in the same published package and only in general terms by subject. The library catalog explicitly identifies these relationships through controlled vocabulary and cross references. In sum, with our current beliefs we do not have a framework that allows consideration of partial authority control to be a reasonable approach.
Why then pursue exploration of partial authority control? For two reasons. It is a reality within local catalogs, consortial union catalogs and national databases, It is also doubtful that complete authority control can ever be achieved within library catalogs or between library catalogs and other databases. If this latter statement is true, then we have to explore whether we can make better decisions in support of effective information retrieval if partial authority control occurs through design and not as a matter of default.
The following list, which is intended to be illustrative not comprehensive, focuses on questions to help answer whether designing and implementing partial authority control would be cost effective while still supporting effective information retrieval from library catalogs.
This is not nearly as radical as it sounds. We are in a situation now where there is only partial authority control and users do navigate successfully among variant forms. Wanting to know whether Ohio State owned a title published by Essen University, I searched the name of the editor - Ahmed H. Helal - in our local catalog and found two sets of entries - some under Helal, Ahmed H., some under Helal, A.H.(Ahmed H.). (This occurred not because our local authority control was remiss but rather because we have loaded ERIC document bibliographic records, and on those records, the headings are not established in the same forms as they are in the LC Name Authority File.) I found what I wanted without much difficulty. We know users have trouble matching their subject terms to library vocabulary but they can and do browse through small sets retrieved via keywords (Meadow 1995, 504).
We could continue as we are, striving toward complete control over personal name headings but it is doubtful we can achieve this goal. In that context, forging ahead only in the direction of total control may unnecessarily constrain the scope and outcome of future dialogues. Of course, we do modify authority control practices to enhance results and increase productivity; indeed this conference is prima facie evidence of the best of that, but the changes we make are still bounded by the goal of comprehensive recall in response to personal names used as search keys and have not yet enabled us to implement complete control over personal names.
The library catalog adds value through the organization and framework imposed on the bibliographic surrogates. It is to everyone's benefit to put our efforts where they count the most. In regard to access standards, this means establishing personal names where doing so contributes most to effective/enhanced retrieval, and not as a numerical input of names included in an authority file.
Meadow, Charles T. "A Study of User Performance and Attitudes with Information Retrieval Interfaces." JASIS 46:490-505 (August 1995).
Younger, Jennifer A. "After Cutter: Authority Control in the Twenty-first Century." LRTS 39: 133-141, (April 1995).
Younger, Jennifer A. "Reframing the Authority Control Debate," IN The Future is Now: Reconciling Change and Continuity in Authority Control, Dublin, OH: OCLC, 1995.
Return to Proceedings Home Page
Advanced Search | Careers at OCLC | Feedback | Privacy Policy | ISO 9001 Certificate | ©2003 OCLC Online Computer Library Center, Inc. |