Summit-Quality control suggestions

Quality control suggestions

"Quality control" for Summit depends on quality control at local sites, since there is no means of changing, adding, or deleting records directly in the union catalog. If a record is improved for Summit, then it's improved for the local site as well! Title filing indicators, record contribution codes, spelling, mis-tagged variable fields, and other common problems can have big impacts in the local and union catalog, and prevent patrons from having access to information. When patrons don't see the title "Residential energy usage comparison" in one catalog, they don't see it in the other, either.

The following suggestions have been gathered from several sources. More ideas? Please send them to us to share.

Barcodes

Identify item records for materials that can circulate and that don't have valid barcode fields. Items without barcodes, or without valid barcodes, cause problems and delays for local as well as INN-Reach circulation.

Missing barcode field. Look for item records limited to certain criteria, or excluding certain criteria. For example, look for item records with a status code "-" (eligible for Summit Borrowing). If you attach item records prior to processing and barcoding the item, and you do not use a specific status code other than "-" you may need to refine the criteria. (Some libraries use "p" for "in process".)
Invalid, or "bad" barcodes. Compile a list of item records with barcode fields, then sort the list by barcode number. Or, compile a list of item records with barcode number < [fill-in-the-blank] or > [fill-in-the-blank]; you're looking for numbers that are not in a valid sequence.

001 fields (OCLC or system number) with inappropriate data

Identify OCLC numbers with inappropriate data that prevents matching and overlay, and other data mis-tagged as an OCLC number. Look for prefixes and other alpha characters before numbers.

Search in OPAC for OCLC# "a" and see what the system displays on a browse screen.
Do Boolean searches, such as OCLC# > 999999, OCLC# >= [greater than or equal to] a, or OCLC# < 0 [less than zero] [with or without another criteria]
look for "ocm" prefix before OCLC numbers (this would prevent matching with same record contributed from other institutions)

Initial articles in titles

Look for titles not findable (i.e. not indexed) by the first significant word. These may be "titles" from many fields: 245, 222, 229, 730, 740, 130, 440, 830, 840, etc. Running a lot of Boolean searches is very cumbersome, even if you search for any "t" tagged field instead of using marc tags; to find "titles beginning with German "Die" you will do a search that has a second variable, so that you add LANG ger and LANG dut -- and so forth. Another technique is to simply do title searches in the OPAC and record or print to e-mail what you find. Initial articles in English and other languages.

BCode3

Unless your institution has a special re-mapping table to convert BCode3 values, be sure that you are not accidentally using any codes that have unintended consequences in the union catalog. You shouldn't have any occurrences of these codes: g, l, and o. ["g" or "o" will cause the union catalog to suppress local holdings and instead display the message "Summit Resource"]. Also be careful how "z" is used. BCode3 Policy and Coding. Local BCode3 codes and Summit mapping

VOL field in item records

Look for unusual information in VOL fields. Caution: in some systems, the "VOL" field that is sent to Summit is labeled "YEAR" or "CHRONO" or other in local catalog.

Catalog Date

Identify records that are (1) already cataloged but don't appear to be, and (2) brief records that have a CAT DATE so they appear to be cataloged. Use two or more criteria. Example for each search:

(1) CAT DATE = - - AND [something you'd find only in cataloged record, such as a certain BCode3 value]
(2) CAT DATE > - - AND [a characteristic of brief records]

CIP records not completed

Look for CIP records that didn't get filled in as the item was cataloged. For example, compile a Boolean review file looking for the "empty" description yet the record has a catalog date:

MARC tag 300 has p. cm.

and

CAT DATE > 01-01-1901

More ideas:

University of Puget Sound(Peggy Firman, 5-23-00)

Article search for:: A, An, The, El, Los, Las, Los, Las, La, Le, Les, Ein, Eine, Der, Die, Das
Link check for all 856's: (we use linkbot, and check all errors manually)
Wrong code because of bib notification: (Bib) loc=zzzzz and cat date > 01/01/1990 and gov pub ~ f
On-the-fly cataloging: (Bib) cat date > xx/xx/xxxx and (item) imessage = f
or (Item) imessage = f
Marcive brief records on the system Muniv check for inclusion.: (Bib) loc=muniv. List all authors (a and b). Check for tracings for faculty. Insure all faculty still at university
Examine each call number index and clean them out: (Done manually, not through list) c=LC v=Wadocs l=SuDocs j= Accession
Clean up zzzzz in multi records: (Bib) loc=zzzzz and loc=multi
List of serials without 222,229's: Blvl=s and Check on items received but not yet cataloged (Bib) Cat date =blank and bsuppr ~ n and (order) rdate w xx/xx/xxxx & xx/xx/xxxx
Cataloged titles with no items (Eliminating federal documents because our Marcive records are loaded with cat dates): (Bib) Cat date>xx/xx/xxxx and gov pub ~f and bsupp=- and link does not exist to item
Marcive records with no items attached: (Bib) Cat date > xx/xx/xxxx and gov pub = f and bsupp=- and link does not exist to item
Ghost bibs: (Bib) Cat date = blank and bsupp=- and gov pub ~f and link does not exist to item and link does not exist to order

from Brown University (Howard Pasternak 12 Nov 2001)

Run statistical reports on bibs and items.  Note the following:
* Bib locations used in items
* Item locations used in bibs
* Invalid fixed fields.  While it is not possible to input these
directly, it is possible to create a workform template with invalid
values and to propagate records this way.  It is also possible to delete
the value from the tables for a code which is still active.
* Filing indicators

Search the database for the following:
* Bibs without 245 or title (t) fields
* Records without "c" tags
* Bbis without attachments (i,c,o)
* Bibs without 008 fields
* Serials without the fields needed for your journal title index
* Authority records without a 008 field

from Claremont (Candace Lebel 12 Nov 2001)

Another way I've rooted out odd skip encoding is to run a statistical report based on all OTHER record data elements on all the bib records in the system. The last time I did this I found 1 record encoded as skip 70, another as skip 21, and another as skip 10. I then run a review file to find them and clean as necessary.

from Quinnipiac University (Terry Ballard 12 Nov 2001)

There is one interesting problem that I search out now and again - titles that have the wrong check digit. To see this in action go to a big database like Ohiolink, Orbis or New York
Public and search the word "tion." A title with an incorrect skip digit is lost to titles searches, but not to keyword. The solution is to run a
search like this:
  LANG = eng AND TITLE has 24504|aA
 OR LANG = eng AND TITLE has 24504|aB
 OR LANG = eng AND TITLE has 24504|aC
 
  so on to Z, skipping T, since the word  "the" is likely the only word in English to generate a skip digit of 4.

updated: August 21, 2007