First, the necessary filters (please use these first, and change the decision field from y to n in the records you find):

1. To find the non-book format materials (we can send ONLY books at this point):

        in the FORMAT field, look for NOT b

2. To find multi-volume sets (we can send ONLY single volumes at this point):

        in the DESCRIPTION field, look for *v.* and not 1 v.*

3. To find books with duplicate Orbis keys (supposedly two copies of the same book in SML, neither of which
has an item record, both on the same bibliographic record):

        From the menu screen that appears when you open your file, choose the report "Print Duplicates by Orbis Key"; this will identify the Orbis keys, which you can use to review the records in Orbis. In some cases the records will indicate that there really are two copies of the book in SML, and you may well decide that both should be transferred. In others, each copy will have a different call number, or there will be some other oddness; these cases should be sent to Catalog Management. For now, hold onto them until we set up a clear procedure with Martha Conway.

Second, some other, possibly useful filters and approaches:

1. To find books published after 1989 (In theory, every book acquired since 1989 has an item record, and thus should not show up in our files; in practice, as we've discovered, there are some pretty recent titles appearing in these files. In some cases, the acquisitions (provisional) record and the cataloging record never found each other, so there are two records in Orbis for only one book—the GOOD bib record has no item record attached to it and the BAD (provisional) record has an item record on which the book has been happily circulating. It's worth stressing, lest you fear that a heavily used book will be transferred out to the LSF under these circumstances, that the barcode on the book should be a red flag during the transfer process. In any case, it makes sense to de-select these recent titles):

        in the DT1 field look for >1989

2. To identify reference materials:

        a. In the CONTENTS field, look for *c* or *d* or *e* or *f* or *i* or *r* (this will pull records coded as catalogs, dictionaries, encyclopedias, handbooks, indexes, or directories--see the web page for "The Selectors' Tool" for links to lists of MARC fields and values);     OR
        b. In the SUBJECT field, look for the appropriate terms (e.g., dictionaries )--make sure you truncate on both sides (e.g., *dictionar*)

3. To find series:

        In the SERIES field, use IS NOT NULL, and then sort by that field

4. To find collections of essays or anthologies:

        In the AUTHOR field, use IS NULL
        In the TITLE field, look for *collect* or *antholo* (or whatever the terms are in your relevant languages)
 

Miscellaneous

        There are records in our Access files for books that have item records and have probably circulated. These records turn up because there are two records in Orbis for the same book: one a "good" catalog record WITHOUT an item record (this is the one picked up by the Orbis report for the Access files), and one an acquisitions record WITH an item record. It is not possible in Access to weed all of these out, although filters for post-1989 books will get some of them. Even if some of these slip through, and keep their "yes" decision status, the books themselves will not be transferred to the LSF because the processing staff will put aside any book with a barcode. Eventually Catalog Management hopes to be able to clean these up.

        Also appearing in the Access files are individual volumes in numbered classed-together series. They are recognizable by the volume number at the end of the call number, but I've not yet figured out a way to filter for them. They will not be transferred, however, because the folks who will be pulling the books will be told to leave them on the shelf and not break up the series.
 
 

mkp 5/17/99