LTI Summary of Emory Authorities Processing

I have checked our records for Emory's use of Authority Update Processing, which goes back to 1994.   We have never received a new bibliographic file from the library.  

While I was looking, I decided to put together a summary of the authority control we've done for the library.  In March 1994, LTI authorized 1,063,118 records for the library.  Subsequent batch processing was performed as follows:

June 1994:                     29,015 records
February 1995:     221,823 records (Marcive gov docs)
March 1995:                    20,049 records
July 1996:                     36,331 records
March 2005:                 126,209 records (EEBO)
Total batch:             1,496,545 records

Additionally, I find that from 1995 through April 1998, the library processed 219,378 records through Authority Express, and another 1,379,813 records from May 1998 through May 2011.  Total AEX: 1,599,191.  All records in AEX files were added to the "base" file of records retained from earlier processing.  

Total number of records processed by LTI through 6/3/11:  3,095,736.  



Jeff Sowder: Note that Emory's weekly AEX service is matching LCAF records with a base bib file from Emory Libraries which is 17 years old! We are getting 'errors' on headings that no longer exist in our current database. 

LTI Response to Laura Akerman Questons

1.  What to do about provisional records:  without provisional records, a large portion (perhaps the majority) of our subject headings would not be linked to an authority (and many other headings as well).   LC has started creating authorities for subject/floating subdivision pairs, but since the possible combinations are astronomical...  we'd always need provisionals, if we need them.

While provisional authority records have been loaded and used in certain ways in the library's Sirsi ILS, one should not make the assumption that they must necessarily operate the same way in the new system.  The basic question of the need (or lack thereof) for "provisional" authority records must be addressed to Aleph.  Out of several hundred libraries using LTI's continuing services, only 8 (including Emory) get provisional records from our processing; about half are Sirsi libraries.  None of our clients that use Aleph are in this group.  This suggests strongly that, if some type of brief authority records are required by the ILS, they are being created at the time they are needed by the system itself.  

We need to find out what the consequences for that would be in Aleph, in terms of
   --updating strategies for these headings and other uses we have had for provisionals such as:
  --storage of local notes about a series, name, etc., as well as
   --a visual cue to catalogers looking at the record that the heading is "authorized," needed for authority checking/validation of individual records and reports of unlinked headings). 

Again, questions for Aleph.  However, if the library moves to a more streamlined workflow, one might expect less focus on tasks like updating provisional headings and authority checking/validation by staff.  LTI processing automatically provides reports of headings which have failed to link on $a and, optionally, "partially validated headings" -- headings which link on $a but not on some subsequent subfield.  (These are, overwhelmingly, name/title headings in which the name portion of the heading could be linked but the title portion could not.)

Note that, in most cases, there is nothing "wrong" with headings that appear on the "unlinked headings" report.  That said, any heading which is really wrong will be found there.   Some libraries do nothing with the report; others examine it in detail.   If asked, I generally recommend that someone quickly scan it for the obvious problem headings, correct those headings in the bib file and re-send the revised records through AEX.  Upon request, I can forward to you a more detailed response I sent to a library suggesting a moderate approach to the unlinked headings list from a batch authorization project.

Furthermore, full use of LTI services will relieve the staff of the need to re-search headings for newly issued authority records, as AUP re-authorizes the entire database, checking headings that did not link in earlier processing.  If a match to an authority record can be made, the record is provided. 

Could LTI generate a new batch of provisionals for us if we reset our "history file" with them?  Or, would  we be able to generate "provisional" authority records in Aleph once we've loaded our bibs, based on unlinked headings, and if so how/when that could be accomplished?   Will we continue to receive provisional records with AEX?

Whether or not a library receives provisional authority records from LTI processing is controlled by the library's profile, and is consistent across both the AEX and AUP services.  If a completely new set of authority records is pulled for Emory, a new set of provisional records can also be provided, based on the profile.  

1(2).:  Local notes in authorities (both LC and provisional):  We don't want to manually re-enter all the notes about treatment of series (analyzed or not) and other notes.  LTI's suggestion seems good, if we can do it, but this would be a job for Systems (sifting out provisionals and LC authorities that have our local notes in them) and their time is pretty booked between now and STP.

Alternative that could be investigated if needed: 
go ahead and load the Unicorn authorities in Aleph;
use an Aleph service to search and identify all records that have local notes (assuming we can identify them by tag/subfield);
export those from Aleph;
delete old Unicorn authorities from Aleph,
then load new set of authorities from LTI (based on our re-set History file which could be the bibs extracted for migration?).
This assumes that we do use provisionals... 
Advantage:   Doesn't push the migration schedule, since it would happen after STP
Possible disadvantage:  We'd have to work out ahead of time, the timing on this, and find out if "down time" for Aleph is required.  Would require some Systems assistance.

Other alternative (not as "clean"):
Go ahead and load Unicorn authorities
Run the service to delete authorities that have no headings linked
Not clear if there would be value to load a fresh batch of LC authorities from LTI, to overlay/update existing LC authorities or load "new" if we don't have that record.
Continue receiving provisionals with new AEX
Disadvantages:  still could retain some of the duplicate provisional mess.  Examining test load 2 could give us a picture of how this plays out in Aleph.

LTI has no remarks on the above, as it all seems to concern local decisions about handling of provisional authority records.   It occurs to me, however, that the library needs to confirm that its authority files can be exported in MARC-21 format, as there was a time when Sirsi systems could not do so.  This may have changed, but it would be prudent to confirm prior to designing complex routines using the exported file.

Neither strategy will give us authorities that LTI sent us already, that didn't get loaded, right?

I cannot emphasize too strongly that the library should send LTI a new copy of its existing bibliographic database and receive a completely new set of authority records, based on the headings in that database.  These authority records should be loaded into the new ILS to create the authority file in Aleph.   This is the only way in which the library can be certain that it has all the authority records needed for the headings in its bib database, only those authority records referenced in that file, and that all of the authority records are up-to-date. 

2.   Level 2 (Comprehensive) processing:  I am starting to see the potential advantage to this.  But, have questions:
Will the revised bibs that come back to be re-loaded from this process, get updates to LCSH with floating subdivisions (the bulk of our "provisional" records)? 

In other words, if the base heading changes (e.g. Afro-Americans changed to African Americans and all the subjects built on this term changed, Afro-Americans |x  Employment, etc.), will LTI change the headings? 

If they don't,  the service isn't "automated enough," so I am assuming they do, but didn't see a statement specifically addressing this.

As stated before, the main difference between the old "Level I" Authority Update Processing now used by Emory, and the current comprehensive service (formerly known as "Level II") is the additional file of the library's revised bibliographic records -- with the corrected headings in place -- in each AUP run.  

If they do, what happens with provisional records?  If we don't get updated provisionals from LTI (which is not a service they offered before), would those  headings show as "unlinked" in Aleph?  Would we generate new provisionals in Aleph?  ????

Again, you will need to consult with Aleph about provisional records -- and perhaps revisit how much effort should go into maintaining them.  

I am a little unclear on how the process "Reviewing reports, occasionally backing out an incorrect change, and doing local bib maintenance" suggested by LTI in the last paragraph, will be accomplished in Aleph.  We need to find out more about "backing out an incorrect change". 

Page 9 of my reply shows an example of a general-to-specific name change:
        The old “100” heading is now authorized in auth rec #no2010182442
        Old: Clarke, J.S.
        New: Clarke, J.S.|q(John Stuart)
        n  82026495
My response continues: 

"This example of a “gsar” (general to specific) change is found in the LSA report of each AUP run.  The goal of this section is to advise libraries where bib file maintenance may, in fact be needed.  The notation “The old ‘100’ heading is now authorized in…”  indicates a strong likelihood that the people represented by the personal name headings are now different.   However, the work flow can be adjusted so that the authority records are already loaded and, at most an unneeded one can be deleted rather than searching for new ones to import.   On occasion, the needed authority will not have been given to the library, but, in most cases, it has."

Bib file maintenance would be to revise the bib record(s) and send the revised bib(s) through AEX.  If the authority record for the more general heading is no longer needed, it can be deleted from the library's file and included in an AEX file of LC deletes at some time in the future.  

I think there is a facility in Aleph for reviewing changes and approving or rejecting them before implementing them in authorities (similar to "load for review" in Unicorn, which probably no one but me has ever used!), but am really fuzzy on it (has anyone seen a description in documentation?)

Approving or rejecting authority records or changes one by one -- in any system -- demands an inordinate amount of staff time.   To streamline effectively the workflow for authority control, the presumption must become that the LTI work is correct until proven otherwise, rather than that staff review is necessary prior to using the LTI-created files.  In the rare case when an LTI error is encountered, we encourage libraries to report it to us so that we can revise our processing and, thereby, make the correction for other libraries.  

Additionally, please be aware that, while many local systems (including Aleph) now contain an "automated authority component", such a module is not a substitute for the comprehensive update provided by AUP.  In fact, use of such software can easily, and monumentally, corrupt a clean database.  Full use of Authority Update Processing goes beyond the "automated record changing" capability that is available through local system software.

Our understanding of local system "authority control" modules indicates that these components use as the basis for all changes the presence of the "old" heading as a 4xx field in an authority record. Nothing wrong with that, as far as it goes. There are several problems, however, in relying solely on that approach. For example, there are times when LC does not carry over the "old" heading -- when that happens, the process breaks down. It's likely that there will be many cases in which a link should NOT be made -- LTI blocks thousands of links which have been identified as "false drops".  Changes that affect free-floating subdivisions cannot be made unless an explicit authority record exists for the whole heading, so the work in identifying these cases falls to staff.  There may also be problems in linking the "old" heading because of variations in capitalization, spacing or punctuation -- in either bib headings or in LC authority records (LTI sees hundreds of such errors in LC records each week.... too many even to report all of them.)    I can't tell how identification and deletion of LC-cancelled authority records fits into this procedure, so I can't speak to that issue.   Internal analysis shows that reliance on the LC record alone misses at least one in three changes made by the sophisticated routines built into Authority Update Processing.

In addition, because AUP re-authorizes the entire library database, in that manner  records can be identified which apply to a heading in the bib file and have been newly added by LC.

LTI Response to Emory Authority Questions

LTI RESPONSE TO AUTHORITY CONTROL ISSUES

                                                                                               

(Original in bold)

1.         If we don’t bring in provisional authority records into Aleph (there has to be a several hundred thousand of those at least), does that mean that all of those headings will be unauthorized and/or that LTI will recreate all of those?  Is this an Aleph thing or an LTI thing?  (Maybe Laura A. knows the answer to that.)  This may be more an Aleph thing but we need to bring in provisionals or have a separate database for them – there is much valuable information in local notes, 690’s, in the provisional records.

(LTI response in plain text)

The question of provisional/brief authority records is definitely one for Aleph.  Most of the time, if they are needed, the system will create them “on the fly”, in the absence of an authority record for the heading as the bib records are loaded. It is likely that Emory currently has some provisional authority records that no longer have a corresponding heading in the bib file.   Adding all of the existing provisional records may create problems whether or not the entire file undergoes batch authority processing again.   If there is useful information in some of them, you may want to investigate exporting just those records and retaining them until the new system is up and running.  At that time, you can explore the options either to add the record(s) or move the added fields if a new record now exists for that heading.

2.         (This is related to #3 below.)  As for one of the “number of factors” in Recommendation under Authority Express (AEX), we want to think very carefully about “how much the process of export and import can be automated,” because we have to remember that there are a lot of authority records changed and/or “suggested” (via “LTI sheets”) for which automation isn’t really an option:

My recommendation was mostly concerned with the export of bib records and subsequent return, though to utilize fully the AEX service, loading the revised records is required.   However, one should note that, today’s climate and demand on staff have meant that, for most libraries – especially those with large collections and a shrinking staff – automation is the only option for the bulk of authority control. 

a.       what I call “Good Samaritan” work, e.g. when an author’s authority record changes (e.g. a death date is added), I have students fix all the records for that author (both “official” from OCLC, and provisional from LTI), and which wouldn’t get done (maybe?) if LTI were to only rely on “official” records

Changes such as addition of death date take place in Authority Update Processing, though a single heading sent through AEX will be revised to the correct, full, closed-date heading even if the revised authority record has not yet been distributed to the library.   However, LTI processing does far more than simply match existing 4xx fields in authority records.  All LTI clients using the comprehensive AUP service received revised bib records with closed dates automatically as the authority records were revised.  There was no need to identify affect headings and then search the bib file to add the death date.  This type of change is straightforward and ideal for automation, with little or no chance for error.

b.       differentiating between authors with the same or similar names (and in many cases, LTI suggests a record that is not the one we want). This is not because they send the wrong record but the heading on the bib. record from OCLC did not have the correct version of the name or subject.

All automated authority control processing relies on matching the incoming form of the heading with the 1xx (or 4xx) in a single record in the national-level authority file.   There are rare times that the record that matches is not the correct one, based on the content of the 670.   On the occasion when this is discovered by a library and reported to LTI, we can sometimes make a change to prevent a re-occurrence.  If the incoming heading is flat wrong for the item, there is no recourse except for the library to find and revise.   There is an e-list for personal names that informs members when a heading is found to be erroneous in a master record.   Perhaps this list would be of use in identifying those names that require bib file maintenance. 

3.         (This is related to # 2 above.)  In Current state of the library’s databases, bullet a., they say that “The library has routinely not loaded new and revised authority records provided by AUP.”  Some of those were not “routinely” loaded because:

    a. they were for the wrong author or title

See explanation above.   Additionally the AUP LSA report identifies a common and serious problem in correct linking: when a name heading is made more specific and, at the same time, a new authority record is established for a different person with the same name, but in the identical form as the now-revised heading.  For example:

Authority record:

001  n 2003010661            was originally:          100  $aMason, Malcolm

It was revised to:      100  $aMason, Malcolm $q(Malcolm D.)

At roughly the same time, a new authority record (for a different person)

001  n 2010080977                        was established with         100  $aMason, Malcolm

While it would seem logical that the person creating the new authority record should search and add data to THAT heading, to distinguish from the original person, all too often, as above, the reverse happens:  the original record has the new info added and the 100 in the new authority record is indistinguishable from the other, pre-revision, heading. 

Though automated processing cannot distinguish between the two "Mason, Malcolm"s, the LSA report has a section to draw the library's attention to the possible need for maintenance.   The section is called "general to specific"; this heading is listed as follows:

*gsar General to specific "100" in Auth Rec linked to name/title bib heading

            --The old "100" heading is now authorized in auth rec # n 2010080977

 old: 100 1  aMason, Malcolm

 new: 100 1  aMason, Malcolm|q(Malcolm D.)

 001: n 2003010661 -- 005: 101230

Note that, if the library had the original authority record (n 2003010661), the revised authority record would be provided in the LCN file. Based on the form of the heading present in the bib record, the new authority record would also be pulled and placed in that file.   Maintenance to change the bib record is needed and the library should also re-send the revised bib so that the corrected form is present for the next AUP.  The unneeded authority record may be flagged to be deleted the next time the library sends an ARDEL file.

b.    because we didn’t need them (i.e. no corresponding headings anywhere in the database, either in the authority or bibliographic file)

Some part of this situation is because there have been no updated files sent to LTI for many years.  LTI has been tracking, in AUP, every bib record ever processed for Emory.    Some of the bibs (and, therefore, headings) no longer exist in the library’s database, but LTI is still tracking the authority records.    LTI has an AEX option to delete authority records, but, unless the “base” bibliographic file is refreshed occasionally, the “extra” authority records will be provided again.  Many ILSs can suppress authority records if no match is present in the bib file so they are not visible except to staff.

A post to the LTI-Users e-mail list from a librarian at large university regarding “blind” references

"Blind references" in an online catalog do not result in wild-goose

chases in the same way they might have in card catalogs. Take, for

example. the reference "Ho, Ho see He, He". In a card catalog the

user would have to flip through cards to find out what, if anything,

 was filed under "He, He." Some references would involve opening

new drawers, and depending on the size of the catalog, covering a

bit of ground. However, many online catalogs tell one right away

what to expect, e.g., Ho, Ho see He, He (0 records). Even if the

number of records does not display up-front, the correct form should

only be a click away. Informing users what the authoritative heading

is, and that there is nothing there, actually prevents wild-goose

chases. This is especially true of subject headings, with their

sometimes counter-intuitive forms."

She then noted that some large academic libraries no longer limit their authority files to those that are found in their bib records, but have loaded the entire LC name and subject files. 

In general, however, it seems more efficient to load the authority records in bulk, then create a report of which authority records are not needed and delete them, rather than assigning staff to locate and bring in records one-by-one.    

c.    because we follow the “rule of specificity” (and for example, don’t replace “year and city specific” conference headings with “generic” conference headings)

LTI supports headings with more specificity than those in LC.  For example, when a conference heading has been established without number/date/place, but the all the library’s headings have such additions, the base authority record is provided as cross-references are present on that record.   We have established thousands of specific conferences with number/date/place (as needed) as internal LTI authority records so that they may be considered “valid” and not appear on the report of unlinked headings. 

-           The relationship between 2 and 3 is that we have to be very careful in having LTI “automatically” load and/or update records because there are many instances where loading their “suggested” records would not be correct.  Not sure of the percentage, but I suspect that it’s significant … and probably enough, I suspect, to wreak havoc (or maybe that’s a bit hyperbolic) with the authority database.

On the contrary, I would venture to estimate that the number of incorrect (for whatever reason) records identified (not “suggested”) by LTI would be quite small, if the LTI and Emory versions of the database were re-synchronized by a full re-authorization project.   In over 20 years of authority control processing, there has yet to be a report of a serious problem being introduced into a database by LTI processing.  

4.         Regarding the “don’t catalog” time (starting on Friday afternoons at 3:00), Laura suggested doing this only once a month instead of weekly.  That sounds like a good idea, but how about if we also change the time from Friday at 3:00 p.m. to something like Friday (or Saturday) at midnight instead.  I guess that this is probably our call, not LTI’s.  Yes, this time was established by the EUCLID team in terms of sending the records out and reloading them.  Monthly would work well and save a little time and money; but it needs to be done at least monthly.

Correct – decisions regarding timing of AEX files is up to the library.  Because AEX is totally automated, files may be sent 24 hours a day, 7 days a week.

5.         Regarding Authority Update Processing (AUP), what is the difference between “Level I” (in bullet a.) and the “comprehensive AUP service” (under Recommendation in this section)?  Is it on the order of the way that we (GovDocs) have MARCIVE reload any record from OCLC if the GPO updates it?  Again, automated updating isn’t going to work perfectly because, like it or not, we have and will continue to have hundreds of thousands of provisional records (that presumably wouldn’t get updated when “official” records do). I, too, would like a more detailed explanation of the “comprehensive AUP service”.

The difference in the old level of AUP that Emory uses and the comprehensive level is that, with current AUP processing, the library receives not only new and revised authority records and reports, but also receives its own bibliographic records, with the revised headings in place.   In effect, the library’s entire database is re-authorized with each AUP run, except that, instead of reloading the entire bib file, only the revised records are provided.  Reloading these bibs (based on the unique library bib id number) and overlaying the old version automatically makes the changes in the library’s bibliographic database.    Of course, the records must be loaded for the revisions to be effective.

6.         I suggest that whatever it takes with LTI (and if they offer this as an

option), we need to keep unauthorized headings “in the loop.”  As far as I understand it (from Laura Akerman), once a record is sent to LTI, we can’t send those to them again, even if headings in that record are still unauthorized.  I routinely find unauthorized headings in records that were cataloged years before, and without the option of resending those to LTI, I am required to manually take care of those headings, including proposing local provisional authority records.  Being able to let LTI work on those records (that were somehow “missed” earlier) would be helpful.  So my question is, does the “AUP” service offer that option? 

There is no prohibition on sending a bibliographic record through AEX more than once.  The library has always had that option.  However, each AUP run re-examines each and every bib record previously sent to LTI, and, in that manner, identifies and provides authority records that were not available at the time of initial processing, but which are available now.  Even the old “Level I’ AUP does this, so LTI has provided the library with many new records for previously-unauthorized headings in the files of “new and revised” authority records. 

This is complicated because Bernardo has explained there are various reasons: sometimes a bib. record or several in a weekly load don’t get sent for some reason; occasionally an authority record doesn’t get loaded into EUCLID; so usually LTI thinks they’ve already sent the heading so won’t send it again even if we change the cataloging date; sometimes an authority record gets deleted by mistake, etc.

The issues of not sending bibs or loading authorities is, of course, local.  But you are correct that, once LTI has sent an authority record, it is not re-sent in AEX, though, if it is revised, the revised version is identified and provided in AUP.

7.         Related to number 6 above, our current profile with them is (I believe)

that once they send a “suggestion” on the LTI lists, they will not do that again, even if we haven’t loaded it.  Let’s say, for example, that we lose some of the LTI sheets and those records just don’t get loaded. 

As above, once LTI has sent an authority record, it is not re-sent in AEX, though, if it is revised, the current version is identified and provided in AUP.  (This is true of processing for all libraries and is not profile-dependent.)  Loading the files provided by LTI, rather than relying on manual identification from the report and loading records individually will help minimize this problem.

Is LTI somehow going to monitor our authority database and let us know that we haven’t loaded records they’ve suggested?  I guess that maybe this has more to do with the $90,000 deal, and I guess that it might also be something that we’d have to do on our end. I believe Steve is right and in regard to 6 & 7, the only way to fix this would be to reauthorize all the bib. records.

LTI has no method, or desire, to monitor the library’s authority database, except to make it possible to “re-synch” bibliographic and/or authority files as an added step in any run of Authority Update Processing. 

By the way, as for how Authority records are working in Aleph, the jury

is still out. [rest snipped]

Here are some examples from the LTI Deletes Report of items that illustrate why the Deletes section of the semi-annual LSA report could not become automated. Manual review is necessary to complete the process. Items that have related headings are also affected, since they need to be changed manually as well.

For each item on the Deleted report, we must manually look up what the new (replacement) form is. Often doing more research if the new form isn’t easily provided. Other sections of the report have the new form provided. It would be helpful if the deletes had this information as well (not sure if it is possible to have it added to our profile)

If the authority record has been deleted by LC, the library should follow suit and delete it too.  There is no question that some bib file maintenance is incident to the deletion of authority records.   But much of it is already done by the AUP run and would be automatically done if a file of revised bib records is received and loaded. 

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 111 2  aGPC (Conference)d(2007 :cParis, France)

001: nb2007012524

////// conference… didn’t touch… new form: …GPC (Conference)

This deletion was reported in the 1^st quarter 2010 AUP run.  LTI provided the new authority record no2009039383.   No change to the bib records for the individual conferences was required.  No bib maintenance needed.

-Delete by LC of Auth Rec prev. linked to LCSH bib heading

old: 150    aPeddlers and peddling

001: sh 85099129

//////new forms:  … Peddlers

                              … Peddling

Same quarter.  Heading was split by LC into two separate headings.  LTI used most general (Peddling) for most of the library’s bib headings, as shown in the LSB report for that quarter.  (The change was made in our copy of the file; Level II AUP libraries also received the revised bib records for overlay.) 

The new authority record for “Peddling” (sh2009009537) has been provided to the library, probably before this AUP run.  If the library felt it was worth the time to re-examine all headings related to “Peddlers and peddling”, it could be done based on the LSB reports.  However, since it would not be likely to have an item with the subject “Peddlers” that was not also about Peddling, the bib heading could be accepted “as is”.  Because a search for one of these terms is also likely to pick up the other, doing nothing seems a reasonable decision.

-Delete by LC of Auth Rec prev. linked to LCSH bib heading

old: 150    aFairy talesvFilm and video adaptations

001: sh 92002219

//////new forms:  … Fairy Tales|vFilm adaptations

                              … Fairy Tales|vTelevision adaptations

… some bib records are film and some are television, some could be both. some manual changes needed

This situation is similar to the above, though “Film adaptations” and “Television adaptations” are more separated alphabetically  than “Peddlers” and “Peddling”.  As before, the library received both authority records.   If the distinction between “Film” and “Television” adaptations is found to be critical, bib file maintenance can correct each record.  

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aTu, Jingyi

001: n  81021518

//////new forms:  … Tu, Ching-i, |d 1935-

                   … Tu, Jingyi, |d 1941-              …two different names & years

This heading was listed in the LSA report for 3^rd quarter 2010.  The authority record deleted was for an undifferentiated name heading, per the note on the record for Tu, Jingyi,$d1941.    Because over 6 months have passed between this AUP run and today, I address exactly what the links were here, except to say that, according to our records, the library has received authority records for both of these two people. 

Currently, the heading “Tu, Jingyi” (no date) will not link to any authority record and will be found on the unlinked headings report.

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aParkinson, Brian

001: n  88609799

//////new forms:   … Parkinson, Brian,|c PGCE     

       … Parkinson, Brian J.          

investigation determines that we don’t need both records

Delete in 3^rd quarter 2010.  Again, the authority record deleted was for an undifferentiated name heading.   No new authority record for any  “Parkinson, Brian” was provided in this AUP, and, at this time, none exists.  This heading, when lacking fuller information ($c as above, or $d) is blocked from linking to prevent a false match.   Level II AUP provides an additional report of unlinked headings that can be checked to determine if the heading was unmatched so that, if desired, a search can be conducted for the correct authority record.  For common names, this may be found to be useful.   As far as I can tell, LTI did not provide the record for “Parkinson, Brian$cPGCE,” (nb2001067897) as it does not exist now in the library’s file.  The authority record must have been added locally.

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aSakamoto, Takeshi

001: nr 90027042  …  

///// this was deleted from oclc authority files, and no replacement form given. A browse by name search comes up with three possible choices. Manual review is needed to complete the process.

1     Sakamoto, Takeshi [100]

2     Sakamoto, Takeshi, ǂd 1899-1974 [100]

3     Sakamoto, Takeshi, ǂd 1925- [100]

Though this particular authority record was deleted, there is still an authority record for “Sakamoto, Takeshi” (n 2010043451), which the library was given at some point.   It is not unreasonable to presume that this is the correct record for the heading.

A couple of examples, similar to what Laura sent, where I can’t envision how manual review wouldn’t be necessary:

The old “100” heading is now authorized in auth rec #no2010182442

            Old: Clarke, J.S.

            New: Clarke, J.S.|q(John Stuart)

N82026495

This example of a “gsar” (general to specific) change is found in the LSA report of each AUP run.  The goal of this section is to advise libraries where bib file maintenance may, in fact be needed.  The notation “The old ‘100’ heading is now authorized in…”  indicates a strong likelihood that the people represented by the personal name headings are now different.   However, the work flow can be adjusted so that the authority records are already loaded and, at most an unneeded one can be deleted rather than searching for new ones to import.   On occasion, the needed authority will not have been given to the library, but, in most cases, it has.

Step 1:  check EUCLID to see what topic Clarke, J.S. writes about = 14 books on ground water

Step 2: look at N82026495 and see what that Clarke, J.S. writes about

Step 3: If the Clarke in N82026495 writes about ground water, then replace N83026495 with the new version and no need to look at or bring in #no2010182442

Step 4: if the Clarke in N82026495 doesn’t write about ground water, then look at #no2010182442 to see if that person writes about ground water; if yes, the overlay N82026495 with #no2010182442.

The above it the basic procedure.  Sometimes more is involved: it may be an entirely different Clarke, J. S. that we want; or, for example, some of the 14 books are about ground water and the others are on another topic, so we need BOTH headings.  Then have to manually figure out which heading goes with each of the 14 bibs.

The process of bibliographic maintenance is the same whether or not a revised bib has been imported, except that, in the majority of cases where the heading was corrected by LTI, local staff will no longer be required to make the revision.

Many “gsar” headings are listed with the notation “The old ‘100’ heading no longer exists as an authorized heading”.   This means that data (usually $d) has been added to the shorter heading but the record is still for the same person.

“Authority Update Processing (AUP): Recommendation: The library

should move to the current, comprehensive AUP service.  Semi-

annual processing could be retained, though, given the size of the

database, quarterly runs should be considered.

Files of revised bibliographic records can then be imported to

overlay the existing version, thereby making needed changes with

minimal staff intervention.  The new and revised authority records

should also be loaded; authority records no longer in LC should be

deleted using the LCDEL file, if the ILS allows.

I’d like more detail about these two sentences of this paragraph and

discuss pros and cons.    Otherwise, deleting authority records will require individual searching and deletion based on LCCN.   This is related to what Laura T. sent you; we don’t see how this would work.  If the no longer valid headings were automatically deleted, someone would still need to look at each one and figure out why it was deleted and what the new one(s) should be.

I’d also like us to ask her about what level of service other libraries of our size use.

By a wide margin, LTI clients of over 1 million bib records use comprehensive Level II AUP; those with over 2 million records do so exclusively.    In general, larger libraries find that is time- and cost-effective to accept the work LTI has done as it is reflected in the revised bibliographic records, the corresponding new and revised authority records, and list of deleted authorities.  Reviewing reports,  occasionally backing out an incorrect change, and doing local bib maintenance requires significantly less staff time and effort which can be focused on the areas in which it will be most useful. 

  

LTI Powerpoint Presentation Follow this link to view a slideshow on LTI services

LTI Authority Processing Questions

On April 28, AMS held a meeting with authority control staff to discuss a conference call with LTI regarding new services recommended by Marsha Hunt. We came up with questions for Marsha and will follow up with a conference call in early May. Ann Vidor will schedule the call prior to May 23.

1.     If we don’t bring in provisional authority records into Aleph (there has to be a several hundred thousand of those at least), does that mean that all of those headings will be unauthorized and/or that LTI will recreate all of those?  Is this an Aleph thing or an LTI thing?  (Maybe Laura A. knows the answer to that.)  This may be more an Aleph thing but we need to bring in provisionals or have a separate database for them – there is much valuable information in local notes, 690’s, in the provisional records.

2.    (This is related to #3 below.)  As for one of the “number of factors” in Recommendation under Authority Express (AEX), we want to think very carefully about “how much the process of export and import can be automated,” because we have to remember that there are a lot of authority records changed and/or “suggested” (via “LTI sheets”) for which automation isn’t really an option:

a.    what I call “Good Samaritan” work, e.g. when an author’s authority record changes (e.g. a death date is added), I have students fix all the records for that author (both “official” from OCLC, and provisional from LTI), and which wouldn’t get done (maybe?) if LTI were to only rely on “official” records

b.    differentiating between authors with the same or similar names (and in many cases, LTI suggests a record that is not the one we want). This is not because they send the wrong record but the heading on the bib. record from OCLC did not have the correct version of the name or subject.

3.    (This is related to # 2 above.)  In Current state of the library’s databases, bullet a., they say that “The library has routinely not loaded new and revised authority records provided by AUP.”  Some of those were not “routinely” loaded because:

a.    they were for the wrong author or title

b.    because we didn’t need them (i.e. no corresponding headings anywhere in the database, either in the authority or bibliographic file)

c.    because we follow the “rule of specificity” (and for example, don’t replace “year and city specific” conference headings with “generic” conference headings)

-      The relationship between 2 and 3 is that we have to be very careful in having LTI “automatically” load and/or update records because there are many instances where loading their “suggested” records would not be correct.  Not sure of the percentage, but I suspect that it’s significant … and probably enough, I suspect, to wreak havoc (or maybe that’s a bit hyperbolic) with the authority database.

4.    Regarding the “don’t catalog” time (starting on Friday afternoons at 3:00), Laura suggested doing this only once a month instead of weekly.  That sounds like a good idea, but how about if we also change the time from Friday at 3:00 p.m. to something like Friday (or Saturday) at midnight instead.  I guess that this is probably our call, not LTI’s.  Yes, this time was established by the EUCLID team in terms of sending the records out and reloading them.  Monthly would work well and save a little time and money; but it needs to be done at least monthly.

5.    Regarding Authority Update Processing (AUP), what is the difference between “Level I” (in bullet a.) and the “comprehensive AUP service” (under Recommendation in this section)?  Is it on the order of the way that we (GovDocs) have MARCIVE reload any record from OCLC if the GPO updates it?  Again, automated updating isn’t going to work perfectly because, like it or not, we have and will continue to have hundreds of thousands of provisional records (that presumably wouldn’t get updated when “official” records do). I, too, would like a more detailed explanation of the “comprehensive AUP service”.

6.    I suggest that whatever it takes with LTI (and if they offer this as an option), we need to keep unauthorized headings “in the loop.”  As far as I understand it (from Laura Akerman), once a record is sent to LTI, we can’t send those to them again, even if headings in that record are still unauthorized.  I routinely find unauthorized headings in records that were cataloged years before, and without the option of resending those to LTI, I am required to manually take care of those headings, including proposing local provisional authority records.  Being able to let LTI work on those records (that were somehow “missed” earlier) would be helpful.  So my question is, does the “AUP” service offer that option?  This is complicated because Bernardo has explained there are various reasons: sometimes a bib. record or several in a weekly load don’t get sent for some reason; occasionally an authority record doesn’t get loaded into EUCLID; so usually LTI thinks they’ve already sent the heading so won’t send it again even if we change the cataloging date; sometimes an authority record gets deleted by mistake, etc.

7.    Related to number 6 above, our current profile with them is (I believe) that once they send a “suggestion” on the LTI lists, they will not do that again, even if we haven’t loaded it.  Let’s say, for example, that we lose some of the LTI sheets and those records just don’t get loaded.  Is LTI somehow going to monitor our authority database and let us know that we haven’t loaded records they’ve suggested?  I guess that maybe this has more to do with the $90,000 deal, and I guess that it might also be something that we’d have to do on our end. I believe Steve is right and in regard to 6 & 7, the only way to fix this would be to reauthorize all the bib. records.

That’s about all I can think of right now.  I’ll send more as I think of them.

By the way, as for how Authority records are working in Aleph, the jury is still out.  There are problems, but they’ve been reported.  However, we didn’t get a chance to do a lot of looking at authority records as we would have liked because so many of them didn’t load properly.  But we do know that the Aleph “Check Record” function (which shows “Doc Validation Error” messages and which are related to “Triggers” … sort of) doesn’t seem to be working correctly and/or it’s confusing (and we haven’t figured out exactly how it works … it would help if we could load records).  For example:

-      “Parallel” linked fields (e.g. one 245 in Chinese characters and another parallel 245 in Pinyin) are showing up as errors (“Required 245 field is either missing or duplicated”).  However, linked 245s (one in characters, one in a transliterated “vernacular”) are allowed.

-      Using the “Record check” button in Aleph pops up “Doc Validation Error” messages (which seem to be related to form of the record, i.e. it’s looking for errors in tags, indicators, etc.).  But to get to “authorizing” errors, you have to go to (Cataloging Module)\Edit Actions\Records Triggers.  So we still haven’t found the equivalent to the “Validate headings” function in WorkFlows.

Here are some examples from the LTI Deletes Report of items that illustrate why the Deletes section of the semi-annual LSA report could not become automated. Manual review is necessary to complete the process. Items that have related headings are also affected, since they need to be changed manually as well.

For each item on the Deleted report, we must manually look up what the new (replacement) form is. Often doing more research if the new form isn’t easily provided. Other sections of the report have the new form provided. It would be helpful if the deletes had this information as well (not sure if it is possible to have it added to our profile)

-Laura Trittin

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 111 2  aGPC (Conference)d(2007 :cParis, France)

001: nb2007012524

////// conference… didn’t touch… new form: …GPC (Conference)

-Delete by LC of Auth Rec prev. linked to LCSH bib heading

old: 150    aPeddlers and peddling

001: sh 85099129

//////new forms:  … Peddlers

                  … Peddling

-Delete by LC of Auth Rec prev. linked to LCSH bib heading

old: 150    aFairy talesvFilm and video adaptations

001: sh 92002219

//////new forms:  … Fairy Tales|vFilm adaptations

                  … Fairy Tales|vTelevision adaptations

                                                      … some bib records are film and some are television, some could be both. some manual changes needed

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aTu, Jingyi

001: n  81021518

//////new forms:  … Tu, Ching-i, |d 1935-

                   … Tu, Jingyi, |d 1941-              …two different names & years

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aParkinson, Brian

001: n  88609799

//////new forms:   … Parkinson, Brian,|c PGCE      

                  … Parkinson, Brian J.          investigation determines that we don’t need both records

-Delete by LC of Auth Rec prev. linked to name/title bib heading

old: 100 1  aSakamoto, Takeshi

001: nr 90027042  …   

///// this was deleted from oclc authority files, and no replacement form given. A browse by name search comes up with three possible choices. Manual review is needed to complete the process.

1     Sakamoto, Takeshi [100]

2     Sakamoto, Takeshi, ǂd 1899-1974 [100]

3     Sakamoto, Takeshi, ǂd 1925- [100]

A couple of examples, similar to what Laura sent, where I can’t envision how manual review wouldn’t be necessary:

The old “100” heading is now authorized in auth rec #no2010182442

  Old: Clarke, J.S.

  New: Clarke, J.S.|q(John Stuart)

N82026495

Step 1:  check EUCLID to see what topic Clarke, J.S. writes about = 14 books on ground water

Step 2: look at N82026495 and see what that Clarke, J.S. writes about

Step 3: If the Clarke in N82026495 writes about ground water, then replace N83026495 with the new version and no need to look at or bring in #no2010182442

Step 4: if the Clarke in N82026495 doesn’t write about ground water, then look at #no2010182442 to see if that person writes about ground water; if yes, the overlay N82026495 with #no2010182442.

The above it the basic procedure.  Sometimes more is involved: it may be an entirely different Clarke, J. S. that we want; or, for example, some of the 14 books are about ground water and the others are on another topic, so we need BOTH headings.  Then have to manually figure out which heading goes with each of the 14 bibs.

Authority Update Processing (AUP):

Recommendation: The library should move to the current, comprehensive AUP service.  Semi-annual processing could be retained, though, given the size of the database, quarterly runs should be considered. 

Files of revised bibliographic records can then be imported to overlay the existing version, thereby making needed changes with minimal staff intervention.  The new and revised authority records should also be loaded; authority records no longer in LC should be deleted using the LCDEL file, if the ILS allows.   I’d like more detail about these two sentences of this paragraph and discuss pros and cons.    Otherwise, deleting authority records will require individual searching and deletion based on LCCN.   This is related to what Laura T. sent you; we don’t see how this would work.  If the no longer valid headings were automatically deleted, someone would still need to look at each one and figure out why it was deleted and what the new one(s) should be.

I’d also like us to ask her about what level of service other libraries of our size use.