Skip to main content
  • Subscribe by RSS
  • Ex Libris Knowledge Center

    Browse search when using a base extremely slow for "ECCO" & some other titles

    • Article Type: General
    • Product: Aleph
    • Product Version: 20

    Here is an example if we do a 'browse title search' in the OPAC while limiting to a large base it works fine

    but when I restrict the search to the qrs logical base it will run until timing out

    Some logical bases will work but very slowly and others will work relatively quickly

    The QR record at has "ECCO" in the 490 field:

    4901 |a ECCO

    Looking in the abc01 tab11_acc, we see that the 490 with a first indicator of "1" is not indexed, just first indicator "0":

    4900# SRS -lx68 2
    4900# 9 - TIT -lx68 2
    4900# 9 chi TITC -lx68 2
    4900# 9 jpn TITJ -lx68 2
    4900# 9 kor TITK -lx68 2

    The XY record at , in contrast, has "ECCO" in a 440 field:

    440 0 |a ECCO

    which *is* included in the TIT index for all indicator values.

    You will need to either change the tab11_acc 490 entries to include first indicator "1" or change the QR fields to a form which is indexed.

    "Eighteenth century collections online" is in an "830 0" field, which *is* included in the TIT index for all indicator values.

    The z01_acc_sequence for "Eighteenth century collections online" is 024349399. SQL shows that there are 354,276 doc records connected to this heading:

    abc01@ALEPH20P> select count(*) from z02 where z02_rec_key like '024349399%';
    **** Hit return to continue ****

    As you point out, the "Title begins with" "eighteenth century collections online" also times out in the QR base.

    The "Title words" "eighteenth century collections online" in the QR base gets 80,000 hits. This means that 22% of all the z02's are for QR. But I find that all the *lower-numbered* z02's for z01_acc_sequence 024349399 are for XY: the XY ECCO docs have numbers like 0101nnnnn and 0102nnnnn while the QR ECCO docs have numbers like 0114nnnnn and 0115nnnnn.

    Thus, the first 200,000 z02's the program looks at are *not* for QR, and it times out before finding 30 which *are* for QR.

    Building the z0102 for the abc01qrs base would solve this problem, but (as seen in util h/1/10) there are 968,587 records in the abc01qrs base -- which is 15% of the 6.6 million total (SQL: select count(*) from z00;). A higher percentage than is desirable.

    I think that a better solution is to use the "Title words" option, entering "eighteenth century collections online" in the box. This gives a good result. It has the advantage that the user can enter additional relevant words, such as:

    "eighteenth century collections online" logic

    I think that *no one* wants to see all 80,000 QRS "eighteenth century collections online" records. The user is *always* going to be looking for a subset of the records in this collection.

    • Article last edited: 10/8/2013