INTELLIGENCE COMMUNITY POSITION ON THE IBM PROPOSAL FOR AN AUTOMATIC TRANSLATION FACILITY

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80B01139A000200040005-3
• Release Decision: RIPPUB
• Original Classification: C
• Document Page Count: 6
• Document Creation Date: December 15, 2016
• Document Release Date: April 7, 2004
• Sequence Number: 5
• Publication Date: June 6, 1960
• Content Type: MF

Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
CODIBp5-58
6 June 1960
MEMORANDUM FOR: Committee on Documentation
SUBJECT s Intelligence Community Position on the IBM
Proposal for an Automatic Translation
Facility
to The attached statement has been drafted to get the reaction of
the Intelligence Community to the IBM proposal., and to ascertain to what
extent closer collaboration is possible in the MT field.
2. Further refinements and additionaal points will need to be in-
corporated in the statement before it is final.ized4 but the present
text will suffice to see whether we can go any further.
3,? With reference to paragraph 9 on extent of departmental
financial participation.. I obviously do not expect members to be able
to make specific commitments within the few days given youv It should'
however,, be possible to have indications of willingness to participate'
and on what terms.
I. This will be considered by the CODID at its next meetings
hopefully this week. The Subcommittee on Mechanical Translation
(SCOMT) will again be asked to sit with CODIB for this agenda item,
25X1
Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
 Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
C -O-N-F- -D- ;-N-T-I-A-L
A CODIB-D-5a
F 6 June 1964
T
D
R
UNITED STATES INTELLIGENCE BOARD
COMMITTEE ON DOCUMENTATION
Intelligence Community Position
on the IBM Promo osal for an
Automatic Translation Facility
1. Need for real-time translation
The intelligence community has an urgent need for real-time trans-
lation. Even if a human translation effort were capable of satisfying
the entire volume of Intelligence community translation requirements,
the time lag (in spite of priority scheduling) and cost inherent in the
human effort still would remain prohibitive. The trend toward increas-
ing volume and rates of production of world scientific and technical
literature alone would dictate the need for real-time translation.
2. Status of the U.S. MT program
The majority of the machine translation efforts in the U.S. are
directed toward long-range research goals. Most projects are engaged
25X1
and recently IBM, has an early operational capability been an explicitly
stated objective. Both Georgetown and IBM have demonstrated machine
results indicative of early operational capability.
Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
 Current status of IBM project
The IBM technique is machine and dictionary oriented, that is, a
25X1
Approved Fs Qr -J'e iQD4 63-IG 'f2DP80BO A000200040005-3
special purpose machine with an extremely large-capacity storage has
been developed to accommodate a method (not yet completely formulated)
of translating language by machine. At the present stage of develop-
ment, word-for-word (entry-for-entry) Russian-to-English translation
which, in most cases, is readable but not necessarily accurate has
been accomplished. Programming, in the usual sense of digital machines,
is not required and the entire series of operations effecting trans-
lation is reduced to a table look-up procedure. The current basic
single pass with certain address modifications to accommodate local
Problem areas and proposed solutions
The fact that the approach is basically lexicographic in nature
leaves a number of essentially important language problems unresolved;
e.g., certain local context problems, proper nouns, word order, article
insertion, synthesis of prepositional structure, syntax and semantic
analysis. The accuracy of translation is contingent upon the solution
of these problems. Lack of programming capability which limits or
excludes the range of logical operations required in the solution of
syntax and semantic problems immediately imposes a constraint on the
system's intrinsic power of translation. IBM proposes to modify the
C-O-N-F-I-D-E-N-T-I-A-L
Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
 Approved For Rele~QQ~,~P$Q1139AOOQ0040005-3
basic systems configuration by including a "word analyzer scheme" designed
to resolve the basic problems of syntax and meanings Optimization of
input and output will be accomplished by phasing-in a series of engineer-
(all languages indicated) will improve to and beyond the point of accepta-
bility as a result of these changes cannot be estimated with any degree of
reasonable certainty. A prudent expectation would make it well beyond
the three year period of time stipulated in the proposal.
5. Validity of the IBM technique
The IBM technique in theory is a valid approach to the machine trans-
lation problem; however., because of its present stage of development., its
potential as a practical translation system per se cannot be reasonably
estimated or evaluated. The proposed modification of the basic systems
configuration can only be successful to the extent that it incorporates
the theoretical solutions to the syntactic and semantic problems.
64 Other applications of special-purpose equipment
The large-capacity photoscopic memory and the equipment associated
with it hold forth extremely interesting prospects for language data
processing other than machine translation. The system is especially
adapted to the manipulation of non-numerical data. As such., its use
in information storage and retrieval., automatic abstracting and indexing
25X1
Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3
 Approved For ReleaiW2089*_VQPijP_~ Ref QP,QA J39A00 040005-3
appears to be not only feasible but also distinctly advantageous over
general-purpose digital computers.
7. Relationship of IB 4 proposal mid other 1 programs
The IBM systems after considerable hardware development, now requires
the expenditure of vast funds for a period of intensive research and
development on a method of translating language by machine. The develop-
ment of such a method depends essentially on the success in solving the
semantic and syntactic problems not only by ILM research but also by other
MT groups* A stored-program general-.purpose digital computer allows for a
wider range of sophisticated logical operations which are essential to the
solution of syntactic-semantic problems. With recent and anticipated
advances in storage technique for digital computers, storage will no
longer be a problem. capacity-and-access-tame-wised
8. Degree of support justified
The degree of achievement in MT to date, and the requirements for trans-
lations (including crude translations available at an early date)$ justify
USIB support for the planning and creation of a central machine translation
facility as a service of common concern to the U.S. Government, and eventually
to private organizations as well. The initial goal should be a pilot opera.
tion with production coverage limited both as to languages and disciplines,
but including experimentation in extracting, abstracting and indexing.
USIB support should be augmented year to year, to the extent justified by
an evaluation of results, until a capability has been achieved for trans-
lating all languages and disciplines required by the USIB.
Approved For Release 20 --!-3 4A-RENP74 aei+$9A000200040005-3
 Approved For Release. 2004/05/05 : CIA-RDP80B01139A000200040005-3
C-fl NF-I-t-E-N-T-I-A-L
-5.-
9o Extent of departmental participation
The extent of financial support generally committed by members of
USIB and other departments or agencies of Government in an over-all program,
including the development phase of the central machine translation facility,
should be a function of their responsibility to promote research, as in
the case of the National Science Foundation, or their need for producing
human translations, as in the case of the Central Intelligence Agency,
Specifically, to launch the initial effort in b'Y 1961, the following
organizations agree to seek funds through their established procedures
in the amounts indicated:
State
Defense
Army
Navy CIA
Air Force -Others
JCS
C -O-N F-I-D-E-I1-T-I-A L
Approved For Release 2004/05/05 : CIA-RDP80B01139A000200040005-3