COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Common Business or Military Computer Communication Language
C00013 ENDMK
C⊗;
Common Business or Military Computer Communication Language

	It is increasingly necessary for computer systems to
communicate with each other electronically without human intervention.
For example,  The inventory computer at an airbase should be
able to requisition supplies from the inventory computer at
some general supply base which in turn should be able to deal
with computers belonging to commercial firms.  The full cost-saving
benefits of computerized data bases often substantially depend
on direct communication between computers without going through
humans.  Thus if an inventory control system merely tells a clerk
to type out a purchase order for some supplies and that purchase
order is received by another clerk who types its contents into
a computer, most of the costs and delays remain.  Any desired
human monitoring of simple transactions is best achieved
by having the computer prepare a draft communication for approval
by a human at the appropriate level of authority.
This requires that the communication have a form suitable for
communication between computers that can be automatically translated
into a form that a human can conveniently approve or disapprove or
modify.

	Such inter-computer communication exists today but always
by specific treaty between the two communicators.   For example,
the Internal Revenue Service will accept the information on
W-2 forms on magnetic tape from employers provided it is in a
form prescribed by the Service.  Also the Fleet Data System
exchanges information between ships on what airplanes and ships
are known to the various ships of the fleet, and this information
has a format that was defined when the Fleet Data System was
designed.

	As an alternative to designing the communication at the
same time the computer systems using the information are designed,
we propose to study a %2Common Business Communication Language%1.
This language would provide semantics and syntax for many kinds
of military and commercial business messages.  The language would
be extensible and usable by all kinds of computer systems.  One
would expect that the designers of all kinds of inventory systems,
requisition systems, purchasing systems, reservations systems and
data base systems would implement communication with other systems
using this language.

	Our interest is not in the electronic level of communication,
i.e. we are not proposing yet another effort to standardize modems,
frequencies, error detection and correction, etc.  Instead we aim
at the semantic content of the messages.
Thus we wish to standardize such messages as

	How many F15s do you have, and how many are in working order?

	Please send 500 spare part number 314159265 to arrive
before 19 January by authority of Colonel XYZ.

	My order number 271828 has not arrived.  What is the current
delivery situation?

	What kinds of pencils do you sell, and when could you
deliver 1000 gross number 2 yellow pencils?

	The main problem is not to standardize the grammar of such messages.
In fact we already have a plan for doing that.  Rather the problem
is to standardize the semantics of a wide class of messages so that
programs can be designed to produce them and interpret them properly.

	This immediately confronts us with the problems of finding a
useful subset of natural language and standardizing its semantic
content.  Thus consider the part of the language used to express
military requisitions.  In its simplest form, what is being requisitioned
is a quantity of an item identified by a stock number.  More generally,
substitutions are possible.  A delivery method must be specified, but
it may be partially left open to depend on circumstances.  The authority
for the requisition or purchase must be specified, and the price
stated or requested or a reference to a prior agreement must be
made.  Even this limited class of messages presents difficulty
at our present level of understanding of natural language.

	Indeed the Common Business Communication Language requires the
solution of a different set of problems from those involved in
putting natural language front ends on otherwise conventional
programs.  In the latter case the problem is to say in English
what we already know how to say in %2computerese%1, whereas our
problem is how to say in the most convenient language for a computer
things that computers have heretofore not said at all.  In our opinion,
this semantic problem has greater scientific interest and greater
practical importance than the well-studied syntactic problems.
From both the scientific and the practical point of view, the interesting
thing about natural language is not how its syntax differs from
computer languages but rather the fact that it is possible to
say in natural language things that computer languages can't express
at all.  Thus this proposal itself is an example of something that cannot be
said in computer languages; indeed it is much too complicated
for computers at the present stage of science.

	We believe that the business communication language problem
is at the right level of difficulty from a scientific point of
view and promises practical application.

	There is a tendency for system designers to propose to
solve the communication problem by designing (say) an entire
military business system and design the communication simply
as part of the larger problem.  Such attempts to "conquer
the world" are doomed to failure.
They require getting everyone to switch to a fixed
system at some definite time in the future.  This locks up the
technology and often disrupts successful local solutions of
local problems.  A common business communication language opens
up the possibility of modularizing the overall system.  Different
companies or different parts of government can proceed at different
rates as their present hardware and software become obsolete.  When
they need to procure new hardware or software, they can do so
according to the best buys currently available using newly developed
techniques.  They are not bound to someone's grand plan of ten
years ago.

	This work is somewhat parallel to the work we are proposing
on formalisms for describing other people's files.
Both are efforts to modularize problems that have hitherto been thought of as
requiring grand systems.  They differ in that one provides a way of
understanding and working directly with other people's data, while the
other is a method of exchanging information at the level of semantics.
The approaches supplement each other, but at present it seems that they
can be developed rather independently.

	Some preliminary thoughts on the Common Business Communication
Language are included as Appendix B of this proposal.  The level
of effort proposed is one Research Associate for the two year
period.  Costs are given in the budget.