XQuery alternative ⁓ map | XSLT 2.0 || 3.0

B B B B Stage four execution:

let $m := c:mM($c)
let $h := c:style($t, $c) => c:hM()
return
let $b :=
<b>
B B B  <m>{ $m - $h }</m>
B B B  <h>{ $h - $m }</h>
B B B  <e>
B B B B B B B  { B 
B B B B B B B  let $e := <e>{ $m + $h }</e>
B B B B B B B  for $r in $e//*:map/*:entry[count(*:value) gt 1]
B B B B B B B  return
B B B B B B B  <r>
B B B B B B B B B B B  <id>{ $r/@key }</id>
B B B B B B B B B B B  {
B B B B B B B B B B B  (
B B B B B B B B B B B  if (map:get($m, $r/@key) eq map:get($e, $r/@key)) then
B B B B B B B B B B B  <mE>{ map:get($m, $r/@key) }</mE>
B B B B B B B B B B B  else (),
B B B B B B B B B B B  if (map:get($h, $r/@key) eq map:get($e, $r/@key)) then
B B B B B B B B B B B  <hE>{ map:get($h, $r/@key) }</hE>
B B B B B B B B B B B  else ()
B B B B B B B B B B B  )
B B B B B B B B B B B  }
B B B B B B B  </r>
B B B B B B B  } B 
B B B  </e>
</b>
return
B B B  xslt-invoke("b.xsl", $b)


B B  during which produces below in-memory $b.

<b>
B B B  <m>
B B B B B B B  <map:map>
B B B B B B B B B B B  <map:entry key="/d.xml">
B B B B B B B B B B B B B B B  <map:value>FA</map:value>
B B B B B B B B B B B  </map:entry>
B B B B B B B  </map:map>
B B B  </m>
B B B  <h>
B B B B B B B  <map:map>
B B B B B B B B B B B  <map:entry key="/d.xml">
B B B B B B B B B B B B B B B  <map:value>FE</map:value>
B B B B B B B B B B B  </map:entry>
B B B B B B B  </map:map>
B B B  </h>
B B B  <e>
B B B B B B B  <r>
B B B B B B B B B B B  <id>/d.xml</id>
B B B B B B B B B B B  <mE>FA</mE>
B B B B B B B B B B B  <hE>FE</hE>
B B B B B B B  </r>
B B B  </e>
</b>

B B B B I then transform the resulted $b with XSLT ( <m>, <h> and <e> is in
different format).

B B B  The dynamic transformation of <m>, <h> and <e>B  in XQuery becomes
cumbersome. How realistic is that I can have XSL equivalent (the
matching/binding aspects before xslt-invoke modules)? I can pass the map node
one-at-a-time to the XSLT processor. Can two map nodes be called each other in
XSLT??
B 
B B B  The logic of the XQY is:
B  <m>B (mechanic model) is the sequence of difference of keys (<id>) in $m
while not present in $hM map. <h> (heuristics classification) vice versa. B 
<e> is the sequence of bias error (<mE> has the upper hand over <hE>), in
essence, match & merge in a map.

B B B  Even better, if this can be done in XSLT way.