Hi,

I brought this up at an in-person user group last year but forgot to come on here and add it to the feature request list.

I think it would be super helpful from a usability perspective to be able to narrow your search based on which metadata values apply to your search results. 

Retail sites do this a lot, and so do libraries. What it means is that if you have 10,000 search results, when you look at the pane with narrowing options, the metadata values that apply to your results are listed (instead of every possible metadata value) and the quantity of responsive results is bracketed behind the value - this allows users to quickly narrow their search in meaningful ways, rather than guessing how search results distribute across metadata values such as date or document type and having to try multiple times to eliminate results from their search.

As an example, let's say I'm looking for a book. I go to the library website and type in a keyword from a title, like "dinosaurs".  on the left side I can see under "Topics" that 950 of my results have "dinosaurs" as a topic, 97 have "paleontology", and 68 have "Tyrannosaurus Rex". This can help guide my narrowing because I am not entering a keyword like "Stegosaurus", which there may not be responsive results for - leading to me having to try and narrow by terms over and over until I hit on the one I want. I can also prioritize narrowing by metadata facets that will seriously impact my search, or give me visibility into where my results cluster around things like certain date intervals, topics, book formats, etc.

This is especially applicable for historical information searches - where terminology used by an organization might have shifted over time, or previous employees were the file creators, leading to uncertainty about what a file might have contained for exact wording or titles.

As an example, I was assisting a staff member with finding a historical agreement. We know it was signed between us and another party, but we're not sure what the agreement was termed, and if it was an agreement, a contract, a MOU, etc. We're not totally sure of the year, and keyword searches are only partially successful because it's an old document and some of the items involved had different names that we don't know. Narrowing down historical results or even agreement files can be very time consuming when you're testing keywords or date restrictions and hoping that the result you're looking for is in there, or when you're crawling through file by file to try and locate the thing you're finding.

Fulltext search is of course helpful in these instances, but if the combined metadata you do know are not enough to narrow your search, being able to use metadata that's present in your search results can be immensely helpful.

If anyone would like further examples, please let me know.