Filtering search results to exclude cases where keyword appears in linked authority record

164 views
Skip to first unread message

Carolyn Sullivan

unread,
Dec 18, 2023, 9:25:22 AM12/18/23
to AtoM Users
Hello all, 

I hope you're all enjoying some fun and relaxation as we approach the winter holidays!  I had a request from one of my users.  She noticed that when you search 'archival descriptions' for a keyword, it returns both archival descriptions containing the keyword, AND archival descriptions LINKED to an authority record containing that keyword.

For example, if you search archival descriptions for 'Grenada' in our repository, including not only just top-level descriptions, you get not only the Arthur Bray fonds with keyword Grenada but also every other copy of the Arthur Bray fonds as well, since 'Grenada' appears in Arthur Bray's authority record.  Is there a way around this, or should we mention it as an issue in a Github ticket?

Thanks so much,
Carolyn.

Dan Gillean

unread,
Dec 18, 2023, 2:55:19 PM12/18/23
to ica-ato...@googlegroups.com
Hi Carolyn, 

What is the full version number of your AtoM instance? I believe that this issue *should* have been resolved by the 2.7.0 release: 
I vaguely remember additional tweaks being made, or at least discussed - I am not sure off the top of my head if the Maintainers have made any further refinements since. 


Cheers,

Dan Gillean, MAS, MLIS
AtoM Program Manager
Artefactual Systems, Inc.
604-527-2056
@accesstomemory
he / him


--
You received this message because you are subscribed to the Google Groups "AtoM Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ica-atom-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ica-atom-users/fe6273b4-6268-4648-a861-6d95cabb276bn%40googlegroups.com.

Lrellis D'erth

unread,
Dec 22, 2023, 6:54:12 AM12/22/23
to ica-ato...@googlegroups.com
Odd.  We are in fact on version 2.7 now.  Is there anything I can do to fix this issue?

Dan Gillean

unread,
Jan 3, 2024, 8:31:14 AMJan 3
to ica-ato...@googlegroups.com
Hi again Carolyn, 

Sorry for the delay over the holidays. I believe that this would be the part of the code to check: 
IIn this Elasticsearch mapping file, 18nIncludeInAll is a method that determines which fields from other entities will be included in the global search results for the current target entity - in this case, which fields from other tables and entities will be included in the information object (i.e. archival description) results. As you can see in the linked version above from our development branch (qa/2.x), the only repository field we index as part of archival descriptions is the authorized form of name. 

It's possible further changes have been made in the 2.7.x minor releases, but otherwise, any local differences may imply that a previous upgrade process was not performed correctly, or that someone in your institution has made those changes on purpose. 

If you wanted to edit this block and remove some fields, it theoretically shouldn't break anything so long as you restart and reindex after. However, we always recommend making a backup of your data before messing with code, just in case! 

If you do make changes, be sure to do the following afterwards: 
  • Restart Elasticsearch
    • sudo systemctl restart elasticsearch
  • Clear the application cache
    • php symfony cc
  • Restart PHP-FPM
    • sudo systemctl restart php7.4-fpm
  • Reload Nginx (might not be necessary, but might as well)
    • sudo systemctl reload nginx
  • Repopulate the search index. 
    • php symfony search:populate
Let us know if that helps. 

Cheers, 

Dan Gillean, MAS, MLIS
AtoM Program Manager
Artefactual Systems, Inc.
604-527-2056
@accesstomemory
he / him


Tim Hutchinson

unread,
Jan 3, 2024, 5:40:56 PMJan 3
to AtoM Users
Hi Dan,

This is interesting, thanks for the additional details. This thread jumped out since I remembered a previous issue with the linked repository adding noise to search results.

In this case, I noticed the same behaviour relating to the linked creator in our instance. It looks like the issue Carolyn described will be fixed by the commit made in September (i.e., not in a public release yet), to remove inherited creator history.

Tim

Dan Gillean

unread,
Jan 4, 2024, 8:18:23 AMJan 4
to ica-ato...@googlegroups.com
Ahh, thanks for tracking down the specific fix, Tim - I should have remembered that, since it seems that I filed the related bug report! 

Carolyn, it does sound like this issue ticket, and the related commit, should resolve the issue. You can wait for the upcoming 2.8 release for this fix, or use the linked commit below as a guide to make local changes. It's also good to have the commit to reference, since it seems there are a couple more places that require changes than I had initially suggested in my last post! 
Cheers, 

Dan Gillean, MAS, MLIS
AtoM Program Manager
Artefactual Systems, Inc.
604-527-2056
@accesstomemory
he / him

Carolyn Sullivan

unread,
Jan 4, 2024, 3:21:39 PMJan 4
to AtoM Users
Thank you so much!

Right now, our team doesn't have our AtoM instance in a version control system, so I suspect we'll be waiting for version 2.8 for the fix... but if implementation of the version control comes before the 2.8 release, I'll be sure ask you all about this.  Thank you all so much for your hard work trouble-shooting this and for getting back to me about the fix!  I really appreciate all your help and dedication :)

Cheers,
Carolyn.

Reply all
Reply to author
Forward
0 new messages