PLEASE NOTE: This is an archived document! It is of historical interest only and does not necessarily represent current Yale University Library practice. For other archived documents, go to: Archived Cataloging Documentation. For current documentation, go to: Cataloging at Yale.
Our purpose in contracting with OCLC for automated authority control services is to accomplish the following objectives:
To update and correct the headings on a library's bibliographic records, the OCLC software relies upon (1) headings and references in Library of Congress authority records (2) preferred headings from the OLUC (3) a catalog of heading variants derived from the OLUC and (4) sophisticated computer algorithms designed to recognize and correct a variety of errors. Together these features (1) minimize manual review (2) significantly boost the correction rate and (3) enable complex heading changes.
Another advantage of the OCLC service is that it generates transaction records instead of replacement records. The OCLC transaction records modify only the affected fields in a bibliographic record, thereby allowing the database to remain completely dynamic while authority control processing is under way.
OCLC also offers a notification service, by which updates to those authority records that we have already received from OCLC are sent, thereby affording us the opportunity to address our ongoing authority control needs.
Finally, the OCLC automated authority control service provides a number of useful reports, including (1) a report that lists headings that the OCLC software identified as problematic but could not correct and (2) a report that identifies significant changes to LC authority records, such as the addition or deletion of 4xx and 5xx references, so that we can effect the necessary change in our catalog.
In recommending that the Library contract immediately with OCLC to accomplish our first objective -- update and correct headings on bibliographic records in the existing Orbis database, generate associated authority records, eliminate unwanted authority records, and accommodate necessary local authority records -- we identified a number of critical activities. We are here today to report on our progress to date on the following:
At this time our intention is to send all bibliographic records to OCLC for authority control processing except error records, deleted records, suppressed records, provisional records, and selected in-process, minimal-level, and CIP records.
Law records will be extracted separately and sent for OCLC processing at a later date.
We have extracted and provided to OCLC a sample of bibliographic records representing materials in Chinese, Japanese, Korean, Hebrew, Yiddish, Arabic, and Persian. We will analyze the results of the OCLC process to determine how these records ought to be processed by the OCLC software.
We intend to extract and provide to OCLC a sample of bibliographic records representing materials cataloged by Manuscripts and Archives staff. We will analyze the results of the OCLC process to determine how these records ought to be processed by the OCLC software.
The OCLC software processes the following headings:
Personal names (100/400/600/700/800)
Corporate names (110/410/610/710/810)
Topical subjects (LC) (6500)
Geographic names (LC) (6510)
Medical subject headings (6502)
Local subject headings that are tagged as 6xx with a second indicator of 4 will not be processed. Local subject headings that are tagged as 690 and 691 will be processed. If the heading matches an LC heading, it will be retagged as an LC heading (i.e. as a 650 or 651 with a second indicator of 0). If the heading does not match an LC heading, it will be retagged as a local heading (i.e. as a 650 or 651 with a second indicator of 4) and reported to us.
The OCLC authority control service does not yet process meeting name headings, uniform title headings, or the title portion of name/title headings.
Of the 48,318 bibliographic records processed by the OCLC software, 14,182 (29%) bibliographic records contained at least one changed heading. Specifically, the software yielded the following correction transaction records:
4,570 transaction records for personal name headings 1,738 transaction records for corporate name headings 7,175 transaction records for series headings 5,993 transaction records for Library of Congress subject headings 144 transaction records for Medical subject headings
Even though correction transactions are not generated for all bibliographic records, a complete array of associated authority records is identified by the OCLC software. For the 48,318 bibliographic records processed by the OCLC software, a total of 57,329 authority records were generated. Specifically, the software yielded the following:
23,155 authority records for personal name headings 7,292 authority records for corporate name headings 6,059 authority records for series headings 20,823 authority records for Library of Congress subject headings
We do not plan to receive authority records for Medical subject headings.
The OCLC software is able to identify some personal name headings, corporate name headings, Library of Congress subject headings, and Medical subject headings as incorrect, even though it cannot correct them algorithmically. The series subsystem does not identify such headings. We refer to the headings on this report as "The Uncorrectables." The software identified 15 corporate name headings, 40 Library of Congress subject headings, and 2 Medical subject headings as incorrect, even though it was not able to correct them.
Finally, we requested and received from OCLC a 100-heading sample of the
approximately 7,500 Library of Congress subject headings that (1) did not match
an authority record, (2) did not generate a correction transaction, and (3)
were not identified as uncorrectable. We refer to these headings as "The
Untouchables." We have requested a similar sample of the 1,740 personal
name untouchables. A preliminary and not entirely scientific analysis of the
subject untouchables reveals that a large percentage of the headings are in
fact incorrect. Our plans for dealing with the untouchables have not yet been
determined. Clearly it would not be practical to manually review each and
every untouchable, especially those that are identified during the initial
processing of our bibliographic records. On the other hand, because so many of
the untouchables are incorrect headings (as opposed to headings that are
correct but not represented by an authority record), we are compelled to find
out if OCLC has any plans for addressing the untouchables.