Cooperative Cataloging Council

Task Group II--Availability and Distribution of Program Records
Final Report

1. Goals
   • Ensuring that records will be widely available quickly and easily.
   • Reducing barriers created by the multiple database environment.
2. Assumptions
   • Timely access in national bibliographic databases is prime goal. Bibliographic and authority records should all be accessible nationally within 48-72 hours of cataloging. Titles cataloged for the program will be searched in the national database before cataloging, as will headings needing authority records. Delays in dispatching or loading records will lead to duplication and wasted cataloging efforts and may require the development of extensive and expensive de-duping mechanisms, which should be avoided.
   • Access and distribution will be most timely if work is done directly online in the national utilities. Multiple modes of participation--online in the utilities, local systems, including the Library of Congress system, LC/MUMS--should be permitted, but TG II feels that ultimately the most effective program participation will be encouraged by a program built around local systems. Permitting work to be done in local systems will encourage program participation.

     Participating institutions will be able to leverage investments in local systems and catalogers' workstations which can be combined for maximum productivity. Some institutions will participate only if and when local systems can be the means of contributing records.

   • FTP will be the best record exchange mechanism. Tape is too unwieldy and slow to accomplish goals of timeliness. LSP will have to be used short-term for authority records, but FTP should be used within 12-18 months. Authority and bibliographic records will require separate FTP files.
   • LC/MUMS is a local system serving LC. Its file is not the national database. Not all of LC's cataloging records qualify as program records. In a truly successful nccp program, LC's cataloging should be considered a subset, albeit a large subset, of the nccp program.
3. Approach
   • Time-frame for long-term objectives: -- 12-18 months and beyond
   • Consider authority and bibliographic records separately.
4. Authority records
   1. Assumptions
      • Current climate of discussions seems to be leading to the concept of exchanging authority records on an international level and to the creation of an International Anglo- American Authority File.
      • Authority records will be established only after searching against the master file.
      • LSP will continue to be a fact-of-life for short-term, limiting participation by institutions that will not commit to NACO work in LC/MUMS or LSP environment.
   2. Recommendations
      • TG II feels strongly that the strategic relationships inherent in the realization of an international authority file indicate that a single, master authority file should be our goal. TG II recommends that authority records can be created in a number of locations (local systems, LC/MUMS, utilities) but then should be sent directly to the master file. They should then be distributed from that master file.
      • TG II recommends that the central file be located at LC. LC still contributes 70% of authority records, and current distribution mechanisms, when replaced by FTP and augmented by new venues for the creation of authority records, can continue.
   3. Recommended model for authority record creation and exchange:
      • We anticipate that most participants will prefer to create authority records in local library systems, which will send the authority records via FTP to the master file, on a daily basis.
      • Libraries that want to make use of the LC/MUMS option or to create authority records on the utilities could continue to do so.
      • The master file should load incoming FTP files within 24 hours of receipt and then transmit them to any institution or utility which wants them within a further 24 hours.
      • Utilities should aim to load the records within a final 24 hour period.
      • Our overall goal should then be a maximum of 72 hours from the point of original output. More frequent transmission, particularly between the central site and the utilities, and more frequent loading at both ends, should be encouraged.
      • Libraries that cannot output FTP files directly could do their authority work on the utilities, which should then transfer the records to the master file. We do not consider FTP from local systems to utilities to the master file to be an acceptable routing.
   4. Development required
      • LC, national utilities, local systems: ability to FTP authority records in USMarc format between local systems and the master file and between the master file and utilities and other service agencies.
   5. Advantages, Rationale
      • Placement: development of the master file needs a central, internationally recognized body to serve as its host.
      • Operations: configuration mirrors the current LSP distribution channel, but methods of data input and data distribution differ, opening up the program to various modes of participation.
      • Single master file eliminates potential problems inherent in creation of multiple files with competing levels of authoritativeness.

      However authority records are created and distributed, TG II believes that their national availability should not fall behind that of the associated bibliographic records. Catalogers and local systems should not dawdle in the creation of authority records or hold onto them when once created. The creation process should be tightly compacted chronologically and tightly integrated with the distribution process. Daily transmission should be the rule for participants.

5. Bibliographic records
   1. Assumptions
      Short-term:
      • nccp records continue to be created in bibliographic utilities--both with original contributions and by editing pre- existing records--and in MUMS both by LC and non-LC users.
      • nccp records continue to be distributed and exchanged by tape and by FTP, with LC distributing (1) original records it and nccp libraries create in MUMS and (2) pre- existing records LC itself uses for copy cataloging purposes.
      • The bibliographic community needs to transcend issues of "ownership" of bibliographic records.
      • Move to FTP as preferred method of distribution.
      • Direct input on bibliographic utilities makes a record publicly available soonest, but this advantage needs to be balanced with advantages of working in local systems.
   2. Recommendations
      Short-term:
      • Use express mail for nccp tape exchange.
      • Tighten distribution timeframes, with more frequent transfers and shorter lag times.
      • Aim to generate/deliver/load nccp tapes within maximum of 2 weeks.

      Long-term:

      • Phase out tape in favor of FTP.
   3. Recommended Distribution Model--the Bibliographic Utility Model
      • Transfer circuit is (1) between the institution and a utility and (2) between utilities, with no central nccp distribution node.
      • Records created in many locations: on local systems, including LC or non-LC libraries in MUMS, and on utilities.
      • Records sent daily by FTP to a primary utility when cataloging not done "online" in the utility. This includes all records done in local systems, including those done in the LC/MUMS environment.
      • Records exchanged between (among?) utilities on a daily basis.

      * Bibliographic Utility Model--Recommendations

      • Utilities must move expeditiously to negotiate an exchange of all nccp records between (among?) themselves. RLG/OCLC exchange of CJK records can serve as a model.

      * Bibliographic Utility Model--Development required

      • CoopCatCouncil should identify the nccp utility-partners. OCLC and RLG are represented on the Cooperative Cataloging Council. Are there other "core" utilities that we should try to involve in the exchange of records?

        WLN and ISM are the obvious candidates since the addition of these 2 would enable nccp to be extended to most North American libraries and to complement the movement toward an Anglo-American authority file.

      • All utilities participating in nccp program record exchange will develop means (1) to allow their member libraries to upgrade existing records to nccp status and (2) to exchange them. As we see acquisition/vendor/Foreign Marc records appearing in the databases (directly or in ancillary files), the ability to migrate and upgrade these records will become even more crucial.
      • Utilities need to develop the means to accept and load nccp program records according to a hierarchy (detailed below).

      * Advantages

      • Records appear in the national utilities sooner than in other possible distribution modes. While some institutions have indicated that an elapsed time of 2 weeks from creation to appearance in the national utilities is acceptable, TG II prefers a tighter time frame. Since these records will be going only to the national utilities, the utilities will have to perform duplicate detection and resolution on them. They possess the appropriate mechanisms, but a compact chronological scheme helps avoid duplicative cataloging efforts.
      • LC does not have to undertake development work that would be required if bibliographic records were to pass through MUMS or CDS.
      • Program gets subsidized decentrally by participants, and libraries can choose whether to seek utility credits by working directly online.
      • Libraries can set their holdings at the same time as they contribute their records to their primary utility.

      * Disadvantages

      • Inter-utility agreements are necessary.
      • Many distribution paths exist and complex file exchange routes are required.
      • nccp records will not be distributed as widely as would be possible if CDS distributed them, though libraries can access them by joining cooperating utilities.
      • LC has access to national program records only by recourse to a utility.
6. Appearance of nccp records in utility databases
   1. Assumptions
      • We need to define the relationship of nccp records to other records for the same piece in order to give guidance on how program records should be treated relative to these other records. If TG I's concept of continuous improvement or evolution of program records is accepted, then we need further to give guidance on exchange of updated records.
      • Fundamental value lies in fullness of the records and in their easy identification as nccp records.
      • nccp records should be created, distributed and loaded as Encoding Level 'Blank' records.
   2. Recommendations
      • For purposes of display in utility databases a hierarchy of record values should be established:
      • Full LC or nccp records (EncLvl 'Blank') should be accorded the highest level.
      • Second tier of records is LC CIP.
      • Third tier is Full Member records (on OCLC = EncLvl 'I'), NAL, NLC, and NLM.
      • Below that are all other records. We are not attempting to differentiate between less-than-full member records, LC 5 or 7's, UKM, etc.
      • In this hierarchy we propose to exploit NAL, NLM, or NLC records primarily for their specialized fields (e.g., key numbers, call numbers, subject headings).
      • Utilities can utilize program records, especially corrected or reissued records, as their database requirements dictate. Program records should be re- exchanged when upgraded, particularly when specialized fields are added (as above), but utility- specific versions of program records will not be identical since utilities perform manipulations and corrections (e.g., global changes to headings) on records even now on an ongoing basis. No attempt need be made to keep utility-specific versions in synchronization.
      • Verification and upgrade of LC CIP records be open to the nccp mechanisms. OCLC is currently investigating opening up the CIP record to enrichment beyond just the collation. We suggest that this might be an appropriate means to relieve LC and NLM of the burden of verifying and upgrading CIP records.

      Application of our recommendations falls into 2 parts, depending whether they target a 'Master Record' database (e.g., OCLC) or a 'Cluster Record' database (e.g., RLIN).

   3. 'Master Record' databases
      • Retain first Full LC or nccp record. Augment with certain unrepresented tag groups from other records.
      • Overlay less-than-full (e.g., Encoding Level 8, 7, 5, 2, 1, and OCLC K, M, L) records and Full Member records (OCLC 'I'), but retain certain unrepresented tag groups if not present on the nccp record. Do not attempt character string match, or retention of tag groups from records with the larger number of fields within a tag group.
   4. 'Cluster Record' databases
      • Make first Full LC or nccp record the Primary Cluster Member.
      • Make any later LC or nccp record a Secondary Cluster Member.
   5. Development required
      • Utilities must develop means to accept and load records upgraded by nccp libraries according to a hierarchy.
      • OCLC will have to develop authorization level consistent with input of EncLvl 'Blank' records.

   We further recommend that the distribution process for these records incorporate a means of notifying LC when any library has upgraded one of LC's less-than-full records, especially CIPs, Encoding Level 7's and 5's (the latter no longer loaded into OCLC). We suggest that this might be a useful step toward developing a standing notification service on the national utilities. We envision this as a service in which the utilities provide cataloging copy (in batch mode) to libraries that have provided them with provisional records.

Michael Kaplan, Harvard University, Chair, Ichiko Morita, Ohio State University, Carlen Ruschoff, Georgetown University, Marti Scheel, National Library of Medicine, Linda West, Harvard University, CoopCatCouncil Liaison
October 29, 1993