Something I would suggest looking at is a comparison of your config.metabib_field table and what’s built in stock. Depending on the age and upgrade path for a database it may not have one of the title fields in the expected place or at all (id 4 or 5, I don’t remember which off hand) and all of the subsequent rows have their ids offset by 1. Since there have been upgrade scripts after that time that referenced cmd fields > 3 by id I’ve seen systems get strange results if they’re affected. I’ve got some SQL somewhere that can be adapted to fix it if needed. Usually after fixing this you can reset all of the weights to stock and get good results.
Actually, since I’ve got a fresh db handy, run this: select id, field_class, name from config.metabib_field where id between 2 and 6 order by id; id | field_class | name ----+-------------+------------- 2 | title | abbreviated 3 | title | translated 4 | title | alternative 5 | title | uniform 6 | title | proper (5 rows) If you don’t get the results above there’s nothing you can do by adjusting relevancy rankings that will fix it. Jason -- Jason Boyer Senior System Administrator Equinox Open Library Initiative [email protected] +1 (877) Open-ILS (673-6457) https://equinoxOLI.org/ > On May 25, 2022, at 5:33 PM, Josh Stompro via Evergreen-dev > <[email protected]> wrote: > > Michele, I think my weights are already different, so I don't think I'm > running into this. The search is coming up with the correct results, it just > seems like the relevance ranking should be placing the title first, or close > to first since the keyword is in the title. > > I'll keep poking at it, maybe I was just running into search cache issues. I > restarted the open-ils.search, but didn't use the cache buster string when I > was testing. > > Josh > > > > On Wed, May 25, 2022 at 3:55 PM Josh Stompro <[email protected] > <mailto:[email protected]>> wrote: > Michele, thank you very much for pointing that out, that does seem promising. > I'll report back after trying the fix. > Josh > > > On Wed, May 25, 2022 at 3:37 PM Morgan, Michele <[email protected] > <mailto:[email protected]>> wrote: > Hi Josh, > > Have you seen this Launchpad bug? > > https://bugs.launchpad.net/evergreen/+bug/1942467 > <https://bugs.launchpad.net/evergreen/+bug/1942467> > > I'm not sure if it's related, but the term "performing strangely" in your > description made me think of it. > > Just a thought, > Michele > > -- > Michele M. Morgan, Technical Support Analyst > North of Boston Library Exchange, Danvers Massachusetts > [email protected] <mailto:[email protected]> > > > > On Wed, May 25, 2022 at 4:21 PM Josh Stompro via Evergreen-dev > <[email protected] > <mailto:[email protected]>> wrote: > Hello, does anyone know of a guide or have any tips on figuring out why a > relevancy ranked keyword search seems to be performing strangely? > > I'm looking at the docs for the virtual index definitions > https://docs.evergreen-ils.org/3.2/_virtual_index_definitions.html > <https://docs.evergreen-ils.org/3.2/_virtual_index_definitions.html> > > I went into the "All Searchable Fields" entry, and used the manage link to > see the virtual fields. And I see that I cranked up the weight for the "Main > Title" a few years ago. But a keyword search for "Hotspot" is showing titles > with "Hotspot" in the 520 field before those that have it in the title. > > Even though the record with Hotspot in the title also has "Hotspot" in the > 520 field. > > If there is a way to see the search details on how the relevancy ranking is > calculated, maybe I can spot where I'm going wrong. > > I'm wondering if I'm missing a step where I'm supposed to disable the non > virtual metabib fields for the keyword class as search fields, or something > like that? > > Thanks > Josh > > > > > Josh Stompro > IT Director > [email protected] <mailto:[email protected]> | 218-233-3757 > ext. 139 | 218-790-2110 > Lake Agassiz Regional Library > 118 5th ST S > Moorhead MN 56560 > www.larl.org <http://www.larl.org/> > Our mission is to enrich lives and strengthen communities. > _______________________________________________ > Evergreen-dev mailing list > [email protected] > <mailto:[email protected]> > http://list.evergreen-ils.org/cgi-bin/mailman/listinfo/evergreen-dev > <http://list.evergreen-ils.org/cgi-bin/mailman/listinfo/evergreen-dev> > _______________________________________________ > Evergreen-dev mailing list > [email protected] > http://list.evergreen-ils.org/cgi-bin/mailman/listinfo/evergreen-dev
_______________________________________________ Evergreen-dev mailing list [email protected] http://list.evergreen-ils.org/cgi-bin/mailman/listinfo/evergreen-dev
