RETROSPECTIVE CONVERSION: A PROGRESS REPORT
Presented to the Library Management Council on 6 June 2001
by

Martha O'Hara Conway, Catalog Management Librarian

The following table shows the status of the retrospective conversion projects that together constitute our current effort. For those projects that have been completed, the number in the RECORDS column and in the DONE column reflect the actual number of records processed. For those projects that are still underway, the number in the RECORDS column reflects our current best estimate of the number of records requiring processing; the number in the DONE column represents records processed to date; and the number in the TO DO column represents the difference between the two.

PROJECT

RECORDS

DONE

TO DO

Arts

146,238

146,238

100%

0

0%

Divinity

120,693

120,693

100%

0

0%

East Asian

179,827

45,347

25%

134,480

75%

Gov. Docs.

243,173

243,173

100%

0

0%

Hebraica

30,000

0

0%

30,000

100%

Lewis Walpole

16,770

16,770

100%

0

0%

Medical

142,216

142,216

100%

0

0%

Near East

32,000

15,431

48%

16,569

52%

Official

1,739,300

998,840

57%

740,460

43%

Serials

175,000

69,241

40%

105,759

60%

TOTAL

2,825,217

1,797,949

64%

1,027,268

36%

Official Catalog

Between 1 January and 31 May, OCLC staff processed approximately 371,634 records from the Official Catalog. OCLC estimated that they would process 407,287 records. Actual production was about 9% lower than projected. Of the approximately 371,634 records processed, 298,702 (80%) were converted by OCLC, 55,081 (15%) were found to have been converted already, and 17,851 (5%) will have to be converted locally. The following table shows how many of the records that have been processed to date from the Official Catalog were converted by OCLC, were found to have been converted already, and will have to be converted locally.

Records to be processed (estimated)

1,739,300

Records converted: Orbis copy added

73,937

7%

Records converted: WorldCat updated

570,578

57%

Records converted: WorldCat created

108,396

11%

Records converted: New record keyed

43,039

4%

Records processed: already converted (ORBIS)

159,069

16%

Records processed: to be converted locally

43,821

4%

Records processed to date (actual)

998,840

57%

As of the end of May, OCLC has processed approximately 998,840 records from the Official Catalog. This work represents approximately 4,563 of a total of 7,942 boxes of cards that make up the Official Catalog. These figures suggest that there are approximately 219 cards requiring processing in each of the 7,942 boxes that make up the Official Catalog, which in turn suggests that there are approximately 1,739,298 records requiring processing in the Official Catalog. This number is very much in keeping with our original estimate of the number of records requiring processing from the Official Catalog, which was 1,815,000. We know that OCLC has processed somewhere between 55% and 57% of the records requiring processing in the Official Catalog.

Serials Catalog

Between 1 January and 31 May OCLC staff processed approximately 50,325 records from the Serials Catalog. OCLC estimated that they would process 47,902 records. Actual production was about 5% higher than projected. Of the approximately 50,325 records processed, 29,851 (59%) were converted by OCLC, 8,476 (17%) were found to have been converted already, and 11,998 (24%) will have to be converted locally. The following table shows how many of the records that have been processed to date from the Serials Catalog were converted by OCLC, were found to have been converted already, and will have to be converted locally.

Records to be processed (estimated)

175,000

Records converted: Orbis copy added

7,255

10%

Records converted: WorldCat updated

29,681

43%

Records converted: WorldCat created

6,716

10%

Records converted: New record keyed

363

1%

Records processed: already converted (ORBIS)

11,390

16%

Records processed: to be converted locally

13,836

20%

Records processed to date (actual)

69,241

40%

As of the end of May, OCLC has processed approximately 69,241 records from the Serials Catalog. Of those, 37,778 (55%) represent monographs and 31,463 (45%) represent serials. This work represents approximately 245 of a total of 1,118 boxes of cards that make up the Serials Catalog.

These figures suggest that there are approximately 128 cards representing serials that require processing in each of the 1,118 boxes that make up the Serials Catalog, which in turn suggests that the total number of records representing serials requiring processing in the Serials Catalog is approximately 143,574. A recent analysis of the monographs processed to date from the Serials Catalog and the resulting change in instructions to OCLC regarding the processing of those records representing monographs both of which are described below -- enables us to estimate that OCLC will process an additional 26,000 or so records representing monographs from the Serials Catalog. We very roughly estimated that the total number of records requiring processing in the Serials Catalog is 175,000. That estimate has proven only recently to be a very good one.

Somewhere between 10 and 20 percent of the cards requiring processing in the Serials Catalog are cards representing serials that are in fact monographic series, meaning that the card provides author, title, imprint, and volume numbering information for the individual monographs in the series. About 76% of the cards representing monographic series are marked "Analyzed," which means there are cards for the monographs in the Official Catalog. For the 24% of the cards representing monographic series that are not marked "Analyzed," cards for the monographs are not in the Official Catalog.

More than half of the records processed to date in the Serials Catalog project are records representing monographs because we instructed OCLC to process the monographs that are listed on all cards representing monographic series -- those that are marked "Analyzed" and those that are not. "Process" in this case means "search for a matching record and provide it to Yale if found." Early in April we instructed OCLC to (1) stop processing records for monographs from cards in the Serials Catalog that are marked "Analyzed" and (2) continue processing records for monographs from cards in the Serials Catalog that are not marked "Analyzed." This should significantly reduce the number of records representing monographs that are processed from the Serials Catalog. We will get the records for the monographs in the "Analyzed" monographic series from the Official Catalog project. We can expect that OCLC will provide us with records for about half of the monographs in the monographic series that are not "Analyzed." Records will have to be created locally -- most likely with the item in hand (not from the card) -- for those monographs for which OCLC will not find a matching record.

Both of these decisions, which are based upon the advice of the Retrospective Conversion Advisory Committee, represent "good news" regarding the Serials Catalog project and should increase our confidence that OCLC can and will complete the Serials Catalog project on time and within budget.

East Asian

Approximately 29,827 records representing monographs were converted in-house between October 1996 and December 1999. Since July 2000, OCLC has converted about 15,520 of the 150,000 records representing monographs that they will convert by the end of June 2003. The records will contain vernacular characters and romanized information, and they will be loaded into ORBIS and available via WorldCat (the OCLC database) and RLIN. The number of records that OCLC has converted to date is smaller than anticipated because (1) many of the OCLC operators working on our project are not yet working independently and (2) OCLC has not yet completed the conversion (from Wade Giles to Pinyin) of the records that represent material in Chinese that reside in the WorldCat database. Our project manager assures me that the project will be on track by October and that her staff will process approximately 6,000 records every month thereafter, which should mean that they will finish the project by the contract deadline, which is June 30, 2003.

Hebraica

In February, we contracted with Library Associates Incorporated for the retrospective conversion of approximately 30,000 records representing monographs in Hebrew and Yiddish. The work will be accomplished in RLIN and should be complete by the end of August 2002. The records will contain vernacular characters and romanized information, and they will be loaded into ORBIS and available in RLIN. This project, too, is characterized by a slower-than-anticipated start-up phase and so there is not yet much to report in terms of number of records converted to date. We are working closely with the folks at Library Associates at this time to make sure that the records are converted according to the specifications that we have provided to them.

Near East

Approximately 15,431 records representing monographs in Arabic, Persian, Ottoman Turkish, and a handful of older languages such as Syriac have been converted in-house since September 1999. We estimate that the total number of records requiring conversion is approximately 32,000 and that we will complete the conversion of these records in October 2002. Like our other records representing material in the JACKPHY languages, the records will contain vernacular characters and romanized information, and they will be loaded into ORBIS and available in RLIN.

ORBIS Loads

We have since early 1999 loaded into ORBIS approximately 514,096 records from the Official, Medical, Serials, and East Asian projects. Of these 96,997 (19%) replaced an existing provisional-level record and 415,932 (81%) were added as new records. We have not loaded any records from the Official Catalog or Serials Catalog projects since January, at which time we loaded records updated or created for us by OCLC as of November 16. The approximately 350,000 records that have been updated or created for us since then are being held by OCLC and will go through authority control processing before they are loaded into ORBIS. At that time we will resume loading the records from these projects on a twice-monthly basis.

Problem Resolution

We on the Catalog Management Team staff are working our way through the boxes of cards from the Official Catalog that are returned to us from OCLC and performing a variety of quality assurance and problem resolution activities.

Quality assurance is one of several ways that we make sure that OCLC is converting our catalog records according to the technical specifications that make up the contract between OCLC and Yale. We select 20 unflagged cards from each box and assess the accuracy of the corresponding bibliographic record (in terms of matching and editing or keying), copy holdings record, and (when applicable) volume holdings record.

Problem resolution, however, forms the bulk of our work. Between 1 July 1999 and 31 May 2001, Catalog Management Team staff resolved problems associated with more than 63,000 records that have been flagged for our attention by OCLC staff. In so doing, we have created more than 11,000 new Orbis records, replaced more than 4,000 provisional-level records, added about 3,000 copies to existing records, and relinked more than 17,000 item records.

Usually at least 20 cards in each box are flagged for our attention by OCLC staff. About 70% of the flags are associated with records that have been converted; the remaining flags are associated with records that have not been converted for one of several reasons. Cards representing records that have been converted are flagged for our attention when we need to (1) verify location, call number, and/or copy and/or volume holdings information (2) relink item records and/or (3) bring in to ORBIS an OCLC record that ought to be there already but isn't. Most of these problems can be and are resolved in Catalog Management.

About 30% of the cards that are flagged for our attention are flagged because the record in question has not been converted by OCLC because (1) the card is a "cover card" for a pamphlet collection (2) the card (or a dashed-on item on the card) lacks location and/or call number information (3) the card represents material in a non-Roman alphabet language for which no matching record was found or (4) the card represents material in a unknown location.

The majority of the cards that are flagged because the record has not been converted (1) are "cover cards" for pamphlet collections or (2) represent material in a non-Roman alphabet language for which no matching record was found. We create ORBIS records for the "cover cards" and we will soon begin routing to appropriate YUL cataloging staff photocopies of cards representing material in non-Roman alphabet languages for which no matching record was found. A smaller number of cards are flagged because the card (or a dashed-on item on the card) lacks location and/or call number information or represents material in an unknown location. We are able to create the necessary records for some but not all of these cards, some of which represent material that may or may not still exist, in a location that can not be determined.

 

Top of Page | Retrospective Conversion Home Page

URL:  http://www.library.yale.edu/recon/progress060601.htm 
Questions to Martha Conway <martha.conway@yale.edu>
Last updated: 25 June 2001