Closed karenmajewicz closed 1 year ago
Seeing a lot of Unpermitted parameter: :range
in the application logs here. Will keep troubleshooting, but apparently this logic has shifted...
Wait... the REAL issue here is that my logic for building dateRange values for Solr was bad. I essentially created a bunch of [* TO *]
(infinity) values here for items without a dateRange value, so those hits are just overwhelming results.
I'm updating the code so those documents don't have a dateRange value and I'm reindexing geomgdev.
Okay... after adjusting the indexing logic to guard against arrays of blank entries things seem much improved.
It also worth pointing out our dateRange search in Solr uses the default "intersects" value, so big ranges [1800-2020] are going to appear in a lot of focused range queries, ex. [2000-2001]. https://solr.apache.org/guide/8_11/working-with-dates.html#more-daterangefield-details
@karenmajewicz Let me know what you think.
^ This change and reindex is complete on geodev: https://geodev.btaa.org/?search_field=all_fields&q=&range%5Bgbl_indexYear_im%5D%5Bbegin%5D=2000&range%5Bgbl_indexYear_im%5D%5Bend%5D=2001&commit=Limit
This seems good and makes sense to me. I think that the Contains and Within might cause frustration for users.
I am curious what the possibilities are though. For example, could we set it up so that a user can specify Intersects or Within? Or, develop an algorithm that preferences items Within at the top?
Yeah, we could add some kinda toggle/radio button/control to let the end user choose between the three options: intersects, contains, and within. How they "work" might be hard to explain to an end user. Maybe there's better labels/language we can use?
Also, it might be a bit challenging to implement, as we'd be adding functionality to the blacklight_range_limit
plugin and that plugin is another one of those seldomly maintained Blacklight gems without clear consensus regarding its future.
I'm gonna close this issue now. If you do want to add that query option toggle, let's open another ticket for that work.
Sounds good - We don't need to add that extra functionality now, but it's good to know what is possible.
What should happen:
gbl_dateRange_drsim
What is happening:
I am reviewing Solr's scoring algorithm on this search (https://geo.btaa.org/?f%5Bgbl_resourceClass_sm%5D%5B%5D=Datasets&q=&range%5Bgbl_indexYear_im%5D%5Bbegin%5D=1830&range%5Bgbl_indexYear_im%5D%5Bend%5D=1860&search_field=all_fields&score=true&explain=true), but not seeing helpful information.