Attribute Weighting in Recommendations Criteria w/o a filter value?

I’d like to use attribute weighting in an item-based criterion in a way that makes sense to me but apparently didn’t occur to Adobe devs. The current setup has a weighting % slider, then you choose the attribute, then an operator, then you must enter a static value, so this is really just a leaky filter.

What I want I want from your AI is this:

• Look at the attributes of the current item
• Look at which attribute(s) I’ve selected to weight
• Then look at the catalog items that also have this/these attribute(s)
• Recommend items with rank based first on similarity in these attribute values
• Ideally, you could also apply an operator here to tailor matching vs. anti-matching, e.g., highly recommend items that are more or less different in these attribute values