Cooperative Name Authorities Meeting

The Library of Congress July 13, 1992

Guest Participants: Duane Arenales-NLM, Liz Bishoff-OCLC, Martin Dillon-OCLC, Bill Gosling-Univ. of Michigan, Ed Glazier-RLG, Christa Hoffmann-NLM, Carol Mandel-Columbia Univ., Glenn Patton-OCLC, Lennie Stovel-RLG, Linda West-Harvard

LC Participants: Linda Bartley, Patrick Bernard, John Byrum, Larry Dixson, Ron Goudreau, Kay Guiles, Betty Harris, Judy Henderson, Gloria Hsia, Diane Humes, Sherry Kelley, Suzanne Liggett, Sally McCallum, David Reser, David Smith, Susan Tarr, Sarah Thomas, Susan Vita, Beacher Wiggins, Glen Zimmerman

Introductory remarks were made by Sarah Thomas, Director of Cataloging at LC, followed by an introduction of the participants. Sarah outlined what she felt were the general goals of the meeting, which were:

1. Discuss procedure and policy changes which will improve cooperative authority programs;
2. Discuss the technical underpinnings of the programs which would encourage contributions, and;
3. Discuss governance issues related to cooperative programs.

The background for the discussion of these issues today is based on ongoing discussions here at LC, the presentations and comments made at the LC-NACO Participants Discussion Group at ALA in San Francisco, and specific proposals for change made to the Library from outside libraries (most notably, the Columbia recommendations). Sarah reiterated that although some decisions having no impact beyond those present could be made today by the invited participants, the bulk of the decisions would be in the form of recommendations to take to the larger cooperative cataloging community for comments.

BASIC ASSUMPTIONS

Sarah invited the participants to propose specific objectives for the discussion, although it soon became evident that the objectives needed to be based on certain agreed upon assumptions about authority work and the functions of authority files. Those agreed upon by the group include:

• Discussion should be based on what we envision a national authority file to be 5 years in the future
• All cooperating libraries will be operating in an automated environment
• Catalogers will be working from workstations with multiple file availability
• There will still be a need for authority files and authority control
• The elements defining authority control will still exist (e.g., each entity has a separate record, authority work is performed to permit user collocation, etc.)
• We are dealing primarily with names (whether used as name headings, subject headings, or series); thesauri are not name authority files
• The need for authority records is still based on literary warrant (e.g., do not predict the need for names not needed for use on bibliographic records or as a reference on another authority record)
• Headings on bibliographic records found on OCLC or RLIN bibliographic records have been established using the appropriate rules
• Authority files/records will need to be maintained

BASIC FUNCTIONS OF AUTHORITY FILES

• Provide unique records for separate entities
• Authority files provide for user collocation
• Authority files contain a reference structure and communicate authorized forms, unauthorized forms, and variants
• Authority files provide the basis for automated authority control (e.g., a file against which bibliographic records are matched, for various purposes)
• Authority files/records are a place to record research done to establish or distinguish names

OBJECTIVES

The meeting objectives suggested by the participants based on the above assumptions also elicited many comments and questions which follow.

• To define the function of the "national authority file" and its relationship with the LC authority file (defined for 5 years down the road, see above)
• To increase efficiency of authority creation and control nationwide (through changed procedures, increased use of automation, etc.)
• To determine the use and user of authority records (on a local or national basis)
• To determine how the LC bibliographic file effects authority control
• To determine who is responsible for maintenance and quality control
• To face the economics of authority control -- balance the problems needing to be solved with the reality of budgets
• To determine what types of authors/names benefit from authority control

Considerable discussion focussed on increasing the number of contributed records. It was noted that the authority files on OCLC, RLIN, and at LC are, in effect, mirror files; what about extending contribution to by non-NACO members? Authority control as currently practiced was identified as being time consuming and requiring too much effort -- how can we make it easier for NACO participation and contribution in order to increase the size of the national authority file.

On the other hand, the general explosion of available files (and potential uses of the Internet environment, or even table of contents or article level author information) could result in greatly enlarged authority files. Concerns were raised in regards to a larger universe of authority files and the increased potential for name conflict and file maintenance it would entail.

With this in mind, discussion also was heard on using searching retrieval techniques to alleviate some of the functions of authority files, such as is found in abstracting and indexing databases like MedLine, AGRICOLA and others, which do not worry about name conflicts (using forename initials instead). Will the authority files of such databases be linked to what we have traditionally considered as authority files when both types are available in the same local systems? If not, how will users handle the differences? It was felt by some that the sophisticated front-end user interface needed to link the various authority files probably would not be worth the cost. It was suggested that A&I services may need to start doing authority control much like library catalogs have done, since it was felt that we probably couldn't rely on such searching techniques as keyword retrieval due to the loss of collocation (Chem. Abs. may have already modified its approach).

Discussion turned to the recommendation of Schottlaender (and others) to reduce the universe of authority records by not making authority records for those headings which do not require cross-references. Are some names more "important" than others? If so, how do we determine which ones? What is the subset of authority records which we can reasonably afford to keep under control (keeping in mind that our users will have do learn to deal with both controlled and uncontrolled headings)? Some participants expressed the view that there should be a range of levels of authority records (such as provisional, minimal, full, etc.). Those deemed "less important" may be machine created by pulling in the 1XX and 245 field from the bib. records for a minimum level record with minimal effort and cost.

Other suggestions to limit the subset of the universe of authority records "important" enough for control-- for example, those used frequently, or by limiting to certain types of works (such as literary authors or historical writers). It was decided that for the purposes of this discussion we should confine the universe to those names that go into our local catalogs (with the long range goal of other file relationships).

Discussion then focused on where is this "national authority file" and how is it maintained? It was suggested that the OCLC or RLIN databases (which have LC records as a subset) should be considered the national authority file, except for the purposes of LC catalogers who should only have to search one file (the LC file) before establishing new headings.

Sarah then suggested, within the framework of the preceding general discussion, the meeting should next attempt to make the suggestions more practical and applicable, especially in relation to the following topics: the CONSER governance model, the LSP technical considerations, recommendations made by LC in response to the Columbia modification proposals, and finally, the future.

CONSER GOVERNANCE MODEL

Linda Bartley, CONSER Program Coordinator at the Library of Congress gave a presentation on the CONSER model of governance, which follows:

In her essay in a forthcoming Advances in Librarianship, Carol Mandel (Columbia Univ.) sets out 11 features that distinguish CONSER. Readers of Carol's list will see that I've tucked her points under my list of seven broad characteristics that, together, constitute what I'll call my CONSER model.

1. Mission & Goals

   A mature program, one that has evolved over the course of nearly 20 years. A clarity of purpose (which we revisit often). The mission & goals are set out in the brochure describing the program. Our staple remains bib records for serials, of which name, series/subject authorities are a premier by-product. In addition to using the bib record to construct the authority record, how about the reverse: using the authority record to build and help maintain the accuracy of the bib record? If an item is published in New York, can't we carry in the 260 the rendition of that place as presented in the authority file?

2. Committed membership

   Some new members since retreat in 1986, some concluded. New style members, including CAS, BIOSIS, Faxon and EBSCO. A by-project includes the rest of our work with the abstracting & indexing community. Especially noteworthy are the associate memberships held by those participating in the USNP.

   CONSER members are concerned about quality and quantity. Quality and quantity tension actively acknowledged--stress here the quality. Sampling review. How do we do as much, if not more, with less and still feel respectable?

3. Team management (Governance)

   We're all in this. Yet LC does have several major roles, particularly those that pertain to documentation, training, and communication. As one of our members has said, the presence of other national libraries and NLC tends to "soften" the leading role played by LC.

   We have a rhythm to our business and a good program calendar for regular interaction. Policy in Nov. (including Councilors), Operations in May (including affiliate). In between, At Large in Jan. and June. Sprinkling of task force meetings.

   Glue provided by an Executive Committee, chaired by a rep from a non-national library. At the moment this job is performed gloriously by Bill Gosling (Univ. of Michigan).

4. A commons database (Gary Byrd)

   For those using OCLC directly, it means of course that we can take advantage of all the records created by others. And others can help us maintain the accuracy of the records. We draw from a lot, we give to a lot of others.

   LC and NLC distribute the records--print, microfiche, tape, CD-ROM. Availability of records to the whole beyond OCLC, all of the other commons.

   There are cost incentives in the form of credits for creating and maintaining records.

   The Program budget--generous yet modest, is provided by OCLC. Growth & non-OCLC users is not just a CONSER problem but a problem very well magnified within the CONSER microcosm.

5. Documentation, Training & Communication

   The CONSER Editing Guide (not just a technical manual), and the CONSER Cataloging Manual are both maintained, and a newsletter is published. The CEG especially is used beyond the CONSER program by non-CONSER libraries as a serials cataloging tool. I've got to be able to understand the instructions in the CEG--and I'm far from a cataloger's cataloger.

6. Accommodation (democracy) & Trust

   With some exceptions, a seamless way to do the work of the program--most already use OCLC in some fashion.

   The members practice self-selection in terms of items to be "conserized".

   We share collegiality, sense of belonging, sense of ownership. We are partners in authentication. Early on, means for pronouncing their own work acceptable...and eventually amending LC records.

   Basic premise of accepting what we find unless it is clearly erroneous. Now moving to the next stage of trust--not even attempting to determine if anything wrong. The presumption of near perfection. With the understanding all the while that if we have an active problem, we fix it.

7. Standards & leadership

   We cooperatively create solutions/guidelines. We cooperatively decide on the problems to tackle. We use standards, we endorse standards, we work to create new ones and to keep old ones relevant.

   Change as a constant. Serials! How can a program that deals with items published over time not itself undergo a name change, a change in its style of management, a change in its categories of membership, and various people changes?

Comments to Bartley's presentation:

It was noted that even CONSER (with only 23 libraries) was having trouble looking globally, we will need to work hard to build bridges and integrate existing international authority files. The "common database" issue led to the concern that those institutions doing their cataloging work on local systems must maintain a separate workflow for their CONSER materials which must be input into OCLC or keyed into both systems. It was suggested that the networks and local systems need to look at better ways to import and export records, as this is one of the "unhappy burdens" of cooperative work.

It was also noted that for point 7 (standards and leadership), CONSER is actually a program in serials librarianship, not just cooperative cataloging. It was felt that this provides a programmatic model for NACO rather than its current operational mode, allowing participants to deal more with policies and issues affecting authority control. Some suggested that NACO has surpassed in areas, such as the shared database and having a freer electronic environment while maintaining a consistent file; also, the fact that the locis of activity has not been as important.

It was noted that we will need to share records and responsibility for maintenance as well. Some libraries noted that their local systems were not in sync with the NAF because they can not afford to do the required maintenance which is becoming more and more difficult.

The issue of timeliness was also discussed, noting that batch processing is not consistent with timeliness. It was also noted that timely can sometimes translate to costly, and that we need to decide what we can afford.

Another advantage seen in the CONSER model was self-authentication, noting that changes to serial records reflect changes in the serial, not differences in judgement. Authority records as well should be assumed to be o.k. rather than scrutinized.

One of the problems with adopting the CONSER model is the "sugar daddy" question. OCLC currently funds the infrastructure for CONSER (approx. $250,000). LC already makes a sizeable contribution to cooperative projects in the form of NACO, NCCP and CSCP-- who will pick up the added costs of a CONSER-like governance structure?

LSP TECHNICAL CONSIDERATIONS

Liz Bishoff, OCLC, outlined the LSP concerns now being considered at OCLC. In brief, what is at issue is whether it is cost effective to move LSP to the new Prism architecture, or explore newer technologies of record transfer such as File Transfer Protocol (FTP) over the Internet.

Currently, authority record creation on OCLC is still being done in the old system, and cannot be moved to the new Prism system (which is OCLC's goal, so as not to have to support both systems for the ease of the utility and its member libraries) until a communication protocol to exchange records is developed for Prism. Since LSP is 1980's technology, OCLC would like to consider options. It was noted by some participants that some libraries may actually wish to create their authorities on their own local systems rather than one of the utilities, and that if FTP file transfers were available, local systems could transmit the records to OCLC, or even cut out the "middle man" and transmit directly to LC, making contribution more timely.

Some of the concerns noted about FTP (or other) file transfers included:

1. timeliness -- how often would batches be run, and would it result in minor lapses of the "mirrored database" concept between LC and the utilities;
2. reliability -- if LSP batch "crashes", information can be retrieved, but FTP is "all or nothing";
3. distribution -- what will these changes mean to other systems such as WLN or UTLAS. If the FTP technology is viable, members of other networks not using OCLC or RLIN may be able to participate in the cooperative programs.

Bishoff also discussed name authority creation in the new Prism system, and the possibility of having decentralized authority record contribution involving all OCLC libraries, even those who are not NACO members. These records would have to be marked in some manner to identify them as "un-authenticated" records. It was noted that these records would only be available to OCLC members unless somehow distributed (ending the mirrored authority file status), but would be available for those institutions using the files for post-cataloging authority control. It was noted that it might take some effort to get non-NACO OCLC members up to the point that they could create usable records and develop a common understanding of the authority file formats and rules. This may have implications on the workflows of libraries on having to deal with a new level of authority headings, similar to copy cataloging workflow situations which arise from using either LC or member copy.

It was decided that LC, OCLC, and RLIN need to further discuss the communications protocol issue, and postpone the consideration of involving multiple local systems until later.

PROPOSALS TO IMPROVE COOPERATIVE AUTHORITY FILE BUILDING

John Byrum, Chief of the Regional and Cooperative Cataloging Division at LC, and Suzanne Liggett, Team Leader of Cooperative Cataloging Team I, outlined seven responses to concerns raised by Columbia University dealing with cooperative cataloging programs.

1. Development of Participants' Manual

   "A single generic manual should be developed and made available. The generic manual would: a) be based on working in the bibliographic utilities.., b) be based on working via LSP, and c) include references as necessary to specific section in the LC [RIs & DCMs]."

   RECOMMENDATION: LC should canvass the participants to determine the need and support for a generic manual and, if generally considered desirable, should contract with one or two of the participants to draft the desired manual, taking into account the various tools which NACO and CSCP libraries appear to have prepared for local use. Early on a decisions will be needed on whether the "generic" manual will be system specific, in which case two manuals will have to be produced. Another early-decision question will have to determine maintenance responsibility for the manual(s). (N.B. The proposal calls for procedures appropriate to an LSP environment and the long-term viability of LSP in terms of OCLC operations seems to be questionable.)

2. Base Authority Work on the Shared Database

   "The 'national' file is not the [LC] bibliographic file, it is the OCLC and RLIN databases; either of these files is sufficiently authoritative to serve as the basis for heading establishment."

   RECOMMENDATION: We agree that it is time to make the file truly national and let OCLC, RLIN, or LC be used as deemed appropriate by the cooperating libraries. In fact the RIs themselves define the catalog for purposes of determining fullness of personal name rather generically as "the file against which the cataloging and searching is being done." There might have to be some adjustments to the procedures to indicate the universe being used as the source file in order to prevent headings from being changed back and forth.

3. Adopt a "Whole Book" approach to NACO procedures

   "NACO participants should be able to establish all types of geographic and name headings."

   RECOMMENDATION: LC should appoint a working group of policy specialists and catalogers to investigate simplification of the "division of the world" lists in view of LC's implementation of the WBC model. It might prove useful to include input from the Cooperative Libraries as well. There should be a concerted effort to move as many geographic entities to the names list as possible. Meanwhile, since LC's Cooperative Teams now include both descriptive and subject catalogers, Cooperative Libraries should be authorized to contribute name and geographic headings without consideration of limits on past practices. If problems develop, they can be dealt with after the fact by staff here.

4. Broaden Governance

   CONSER should serve as model for establishment of a steering committee or some other form of governance structure to include participant representation.

   RECOMMENDATION: As a first step, Cooperative Cataloging should meet with Serial Record to discuss the possible adaptation of CONSER governance structure to its authority projects.

5. Develop a more Participant-Oriented Training Program

   NACO libraries should be able to train each other and LC staff should work at member libraries to facilitate training, because the existing method of training new participants is too LC-centric. Training time at LC should be decreased as well.

   RECOMMENDATION: As new participants join the authorities project, they should be given the option of undertaking training at LC or of receiving LC staff to provide the training on site for one or more persons. In either case the training program will be reduced in length, provided a mentor from among already independent NACO participants can be arranged to supplement the introductory LC portion and to assist the new participant's progress by reviewing its authority records before they are contributed. The Participant's Manual called for above would facilitate this approach and as a further measure LC might investigate development of a computer-based tutorial to assist training. In addition, focused-training seminars could be held in conjunction with ALA meetings.

6. Develop Consistent Guidelines

   "Responsibility for revision is widely dispersed at LC ... expectations of revisor-catalogers are widely variable. There is no commonly held view of the appropriate levels of research...Documented guidelines, developed jointly..., would provide an improved, consistent approach toward authority work."

   RECOMMENDATION: Ask participating libraries to confer with the Cooperative Cataloging Team Leaders on any occasion when they feel that they are being asked to undertake research that is appropriately demanding or to make additions or changes to contributed records to which they object.

7. Adapt Rule Interpretations to the Current Environment

   Many rule interpretations and other requirements embody LC-centric file requirements and do not reflect changes brought about by online catalogs.

   RECOMMENDATION: Invite suggestions to improve the RI's and DCM's to make them more congenial to cooperative cataloging from Participating libraries, CC:DA, and the profession at large. Assemble a team of catalogers and policy specialists to review the suggestions and propose their disposition. Review the proposed disposition with the Participating libraries, CC:DA, and the profession-at-large.

Participant Comments on LC Proposals

• Issue 1: Other good manuals exist, so we would be working from an significant base. As far as maintenance is concerned, that would be a governance issue to be decided by the appropriate committee. As to the question of whether a manual should be generic or based on either OCLC or RLIN, it was suggested that this question may be a good one for a participant survey to determine what is needed.
• Issue 2: Some participants are anxious to get started, and would like to know when to get started.
• Issue 5: Whoever trains new members (nearby NACO libraries or LC) will have to coordinate with OCLC or RLIN so that the LSP training can coincide. Some felt that it may still be useful to have some training at LC as well, but not for as long as is currently done.
• Issue 6: The desire for consistent guidelines is coupled with the desire for timely responses from LC to questions and record reviewing. The participants noted that this effort should be truly cooperative, with a freer vision of how we cooperate and allowing for cataloger judgement.

WRAP-UP

It was felt by some that if a CONSER-like system of governance is initiated, many of the other details (such as participant manuals, or guidelines) would fall into place after actions by executive councils and task forces. It was noted that with the changes discussed today, cooperation should be easier because NACO could be made more programmatic, automation advances will make it easier, and economic pressures mean we all need to rely more on each other. In the words of one participant, "It is the right time with the right group of people."

Interest was expressed in the idea to approach CLR for funding of an initial NACO "retreat" similar to the one which developed the CONSER governance structure. Other questions left unanswered include:

1. should we combine NACO and NCCP in these discussions, what about abstracting and indexing databases, or companies like BNA?;
2. a need to address a "national cooperative file" for subjects (we already have a sort of NACO for subjects in CSCP); and,
3. a possible electronic alternative to meetings such as the one today.

Sarah Thomas concluded by saying that LC will continue to focus on the issues raised in the meeting, solicit further input from the participants and the greater library community, and inform the participants of any ongoing discussions.