Domain Knowledge
-----------------

There is little domain lnowledge.  This is sketchy but is just
enough to go fron domian independent utterance rules to speech
acts containing domain dependent language.

Subjects
--------
There are lists of subjects
For each subject list we have sentences for each subject in the list
  (SUBJECT? |medicine|)
  (MEMBER |medicine| |master-subject-list|)

Medicine
--------
There are lists of problems for each subject
For each subject we have sentences for each problem in the list
  (SUBTYPE |medical-problem-name-list| |problem-list|)
  (SUBJECT-OF |medical-problem-name-list| |medicine|)
  (MEMBER |injured| |medical-problem-name-list|) 
  (RELSYM-OF |injured| |INJURED|)

  (PROBLEM? [medical-problem])
  (NAME-LIST? [medical-problem-name-list])
  (MEMBER |injured| [medical-problem-name-list])
  (SUBTYPE [medic] [agent])


  (IMPLIES 
    (AND (TRUSTS ?A) (INFORMED-BY ?A REL1-1) (TRIPLES-of ?A REL1-1))
    (INJURED ?P)

  ; (triples-of ?A2 REL1-1)
  ;   is an abbreviation for the 'list' of properties
  ;   that ?A2 attributes to REL1-1     

  ;  this is domain knowledge bu
  ;    in the absense of reification or existentials
  ;    this is the best we can do!!!
  ; BUT now we have a timing race between U1 and domain knowledge
  ;  ie which is the 'generator' of the symbol REL1-1
  ;  if it's NLP then how does O know about the domain knowledge
  ;  if it's O then how can the NLP system know about it
  ;  or do they share (and with whom?)
  ;  we beg the question