CIA/OCR (SANITIZED) TEST

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80B01139A000200020010-9
• Release Decision: RIPPUB
• Original Classification: C
• Document Page Count: 14
• Document Creation Date: December 22, 2016
• Document Release Date: May 31, 2012
• Sequence Number: 10
• Publication Date: June 30, 1960
• Content Type: MF

Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
C"O" _ - ?
CODIB-D-23/1
30 June 1960
UNITED STATES INTELLIGENCE BOARD
COMMITTEE ON DOCNTATION
MEMORANDUM FOR, USIB Committee on Documentation
SUBJECT: CIA/OCR MINICARD Test
REFERENCE: OODIB-D-23 (8 January 1959)
1. Attached for your information is a summary case history of the limited
test of MINICARD as a substitute for the OCR Intellofax System.
2. Our findings are negative. This conclusion is based only in part on
our findings that the MINICARD system would not enable us to give substantially
superior reference service over that possible with our present system. Wci.ng
very heavily were the present limitations on staff, on space, and on money; these
operating assets have been appreciably reduced since the inception of the MINICARD
project. Moreover, this reduction has occured in the face of an increase in
demand for OCR information services generally, but a relative decline in the
demand for literature searches.
3o The decision not to adopt MINICARD as an operational system in OCR
does not affect in any way the application of this system elsewhere in CIA.
MINICARD has been selected by the CIA Photographic Intelligence Center as a
subsystem of its data handling system. As a consequence of the OCR decision,
MINICARD equipment, spares and supplies will be released to PIC to augment
their proposed MINICARD installation.
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
C -O-N-F-I-D-E-N-T-I-A-L
SUMMARY CASE HISTORY:
LIMITED TEST OF THE MINICARD SYSTEM
AS A SUBSTITUTE FOR THE
CIA/OCR INTELLOFAX SYSTEM
CODIB-D-23/1
30 June 1960
Purpose of the MINICARD Test
1. The objective of the OCD Zn-ow OCR proposal of 25 April 1955 submitted
to the Project Review Committee was,
To conduct in OCD an early and large-scale test of a family of
data handling equipment known as MINICARD9 which is believed capable
of substantially improving CIA's Intellofax System as a principal
instrument in support of intelligence research. MINICARD promises to
contribute improved means for collation of intelligence data, greater
speed and flexibility in the conduct of document searches and economies
in operation, notably spacewise. 1f
Events Leading to the Test Proposal (1955)
2. From its inception in 1947 to 'the time of the MINICARD test proposal,
the storage and retrieval capability of the Intellofax System was increasingly
strained by the flow of information until, by 19559 storage, retrieval and cost
problems were considered urgent. The increased growth of the file had been
accomplished by multiplication of IBM equipment rentals, storage units and
personnel. Analysts' requests at that time for total searches of the seven-year
file amounted to 60% of the requests received.' Compliance with those requests
in categories numbering tens of thousands of cards lengthened search time,
multiplied overlap problems,- and overloaded requesters with insufficiently
refined answers. The possibilities of additional space and personnel ceased,
and the alternative to an improved system was reduction in range, speed, and
quality of Intellofax service.
3. The proposed MINICARD system held prospects of being such an improved
system. MINICARD was said to combine discrete item control, multiple access,
flexibility of electronic searching techniques, and inviolate film storage. It
could combine coded information and document images which were handled separately
by Intellofax. It could ease the critical storage problems with cards and hard-
copy. Much faster retrieval rate was expected. In addition it was anticipated
that MINICARD would "hasten and expand adoption of common data handling procedures
throughout the intelligence community." Six people would comprise the test group;
regular search requests after mid-1954 would be submitted to both NINICARD and
Intellofax and results compared; decision was expected about 1 July 1957 assuming
delivery of equipment about 1 July 1956. Cost reductions would be substantial if
other agencies used MINICARD and exchanged Minicards after processing on a common
basis; savings in machine rentals would be substantial; the major economy would
be space.
C-O-N-F-I-D-E-N-T-I-A-L
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
4. Anticipated results, then, were for (a) a community program for compre-
hensive one-time processing; (b) common community storage and retrieval using a
common code, identical equipment and procedures, and inviolate code record and
document storage; (c) improved reference service, particularly with retrieval
according to subject associations, prompt access (50-759 faster), and essentially
simultaneous processing of overlapping requests; (d) economies in operation.
Estimated equipment cost was
Developments from 1955 to the Test Period (195
5. When the MINICARD project was approved and an order placed in June 1955,
the equipment was in the blueprint stage. Delivery, scheduled for completion in
December 1956, actually occurred in November 1958, with installation completed
in February 1959. The test period, beginning with document selection, ran from
15 January 1959 to March 1960. During the period between the order-date and
delivery, several major modifications were made in the equipment; we were aware
of them but had no legitimate basis for objection since ours was a program appended
to the Eastman Kodak/Air Force development program. The changes did, however,
invalidate the earlier space and personnel estimates. Operating speeds on the
duplicator? sorter, selector and processor were substantially reduced. Also In
the interim, improvements were made in the IBM equipment in the Intellofax System,
particularly with regard to operating speeds (see para. 28). Moreover, techno-
logical developments in the U.S. and elsewhere with various applications to
storage and retrieval of information advanced very rapidly from the time of the
original MINICARD proposal to the beginning of the test.
6. By the time the test phase arrived, the. earlier expectations from
MINICARD had been somewhat modulated. There was some feeling that extreme
miniaturization which eliminated manual access might prove inferior to our
16mm aperture card system. The combination of codes and images in the same
card was being questioned, as was the loss of the bibliographic Intellofax
tape. Finally the changed equipment specification stimulated the belief that
additional purchases would be required at an estimated cost of 3-5 times the
costs in the original order, with an estimated equipment delivery lag of 1-2
years.
7. a t the increased pessimism, the problem remained, requiring
solution.2J V On the one hand, intellofax was a going system and not a
first mechanization approach (hence any conversion must include minor service
disruption, preservation of proven features of the existing system, plus thorough
proof of the replacements); on the other hand, there were no developments in
sight to enable Intellofax to manage 10-20 year indexes with the staff and space
allocated to servicing a 5-year file. MINIC.ARD, though untested, was the only
alternative to EAM (Electrical Accounting Machines) claiming the capability to
handle our task. Not to be overlooked was the fact that at this point, approxi-
mately
had been invested in equipment, supplies and training; another
25X1
was required during the last half of FY 59 for maintenance, repair and
25X1
stocking of spare parts; an over-run claim of above the original contract
25X1
was under review, not to mention the cost of a one year test.
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
-3-
8. Intellofax at this time, as today, allowed for 1-2 day dissemination
of all documents received - a fact often overlooked in considering the total
system; it provided an inviolate (aperture or 35 mm reel) file of documents
and enclosures.; it allowed for manual access by clerical personnel; it included
an index record of subject content of documents. The criticisms of Intellofax
were generally., (a) slowness of dissemination of documents (considering the
total reporting cycle from inception with the collector); (b) incompleteness of
document file (in that poor copy documents., being non-reproducible, were
excluded); (c) slowness in retrieval and lack of precision of results. Since
Intellofax was backed by IBM cards, the additional complaints touched on limited
capacity to record long index terms; slow input and unreliable output, inaccessi-
bility of the cards to the analysts.
9. Still prior to the test, the thought was expressed that subject coding,,
the slowest phase of Intellofax, would be even slower for MINICARD because of
the increased coding complexities; a dual MINICARD scheme (codes and document
images on separate film cards) would reduce equipment processing capability by
500, double the storage space, and create early requirements for additional
equipment; the 60-1 reduction ratio was extreme and would heighten the problems
with poor quality originals.
10. So went the pros and cons. The final recommendation, which was acted
upon affirmatively, was that having already made a substantial investment in
t INICARI) and in spite of the reservations which had developed, OCR should make
'irr its own direct evaluation and proceed with the minimum test as described below
Test Period-. Input Plan
11. The test program was separately staffed and operated to minimize
interference with the regular Intellofax operations. A MINICARD working group
consisting of some of our most experienced people from several OCR divisions
was formed to prepare for input a sample of 20-25,000 raw information reports
containing a normal mixture of document categories by source, format, enclosure
variety and with every kind of problem in subject coding and photography,
including papertypes, inks, size differences,, color differences and varying
legibility; we utilized, on the average, 18 people (average salary) for 25X1
over a year. The test corpus was to be fully processed into (a) Intellofax;
(b) MINICARD, with codes and document images stored together in a single card;
(c) MINICARD, with codes and dcr ument images stored in separate but related
cards. Input was to be accomplished in 8-9 months, depending on Minicoding
techniques then to be perfected.
12. A detailed procedural manual/ was compiled, to be corrected as
experience dictated. It was distributed to Eastman Kodak for information and
comment.7/ It included sections on the equipment., system work flow, machine
operations, MINICARD coding procedures, phrase coding and retrieval procedure.
After 9-12 months of input and retrieval-testing of each of the three approaches
mentioned above, the plan called for projection of performance rates and costs
of each to fu1.1-scale operations, utilizing a minimum five year document collection.
Coding of the corpus was carried out throughout the test for Intellofax and the
?'' separate code/document image MINICARI) approaches. A coding variable which was
0-0-N-F-1-D-E?-N-T-1-A-L
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BOl 139A000200020010-9
1""'C-O-N-F-I-D-E-N-T-IA-L
_4-
not originally proposed was the use of the revised Intelligence Subject Code
for the MINICARD retrieval test and the old form of the ISC for the Intellofax
retrieval phase.
13. In addition to the basic purpose of equipment comparisons and the
secondary operational testing of the revised ISC, it was thought that by-product
information could be obtained on: (a) the role and importance of a source card
file; (b) the utility of a hard-copy document collection; (c) the effect of OCF
policy on "nodexing", i.e., not indexing, certain information.
Output Plan
14. The retrieval phase was to begin in late January 1960 (it began in
February) after completion of the coding, completion of machine input, consul-
tation with other Agency offices, and selection and preparation of the test
questions.. This test phase was to run a month, the first two weeks of which
would involve submittal of 200 questions at the rate of 20 per day. Since
reference service is the justification for all the documentation activity,
including machine operations, its goals were the continually prominent yardstick
of the investigation. The determinants of customer satisfaction to which the
yardstick was applied involved: (a) comprehensiveness, to ensure that all
pertinent items are retrieved; (b) precision, to ensure that all items retrieved
are pertinent; (c) speed, in terms of gross elapsed time. The 200 test questions
Included action (live) requests and simulated requests, suggested by retrieval
test, other OCR, other CIA and non Agency personnel. Also involved would be
requester interviews to define and refine questions and to check customer
satisfaction with content and format.
15. An outline of evaluations to be considered, looked something like'
the following:
a. Comparison of Intellofax and MINICARD for the test corpus and
projection for five-year file
(1) Personnel and training requirements
(2) Machine and supplies requirements
(3) Monetary costs
(4) Quality of reference product
(5) Capacity for normal and crash requests
b. Usefulness of clear text for retrieval
c. Reliability of phrase structure vs. sentence links for retrieval
d. Free vs. bound modifier practice
e. Code rules vs. "word" additions to code outlines
f. Files management procedures
g. Coding practices and tools (including authority files, tag
definitions, coding depth, area code adaptations for subject
modifier use, pagination coding, etc.)
h. Fixed field coding by clerical vs. analysis personnel
i.: Nodex criteria
J. Retention value and purging.
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BOl 139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
statue l0
16. Approximately 15,000+ documents were coded for input and 185 test
questions were processed. For Intellofax the retrieval end product was the
Card-List Camera tape of bibliographic citations; for MINICARD, a first page
print was provided for each document retrieved by the selector. Of the 185
runs made, 120 retrieved specific dents in one or both systems; 65
retrieved no specific documents. Both systems combined retrieved 997 specific
references of with MINICARD obtained 788 or 79% of the total and Intellofax
obtained ' 649 or 65% of the total. MargineGl. or rejected references in MISICARD
were 426, and in Intellofax, 65%. Returns were analyzed according to three
categoriess (a) close match, in which 85% or more of the specific documents
were retrieved by both systems (these represented 19.2% of the 120 analyzed
rime); (b) disparate match, in which 15% or less of the specific documents
vexe retrieved by both systems (36.7% of the 120 runs); (c) divided relevance,
in which 16-814 of the specific: doc uts were retrieved by both ?ystems (44.1%
of the 120 runs).
17. Of the 120 runs which recovered relevant docents, 73 included clear
text in addition to numeric coAesj in 53 of thesee, documents were retrieved by
logical expressions requiring clear text. Among the 23 close-match runs, 13
involved docusents retrieved logical
~ di to runs 28 include"c~ eogical esa~aressionts regegiring c7.ear team? in
quations using clear tent ,?- in 15 of these,
documents were retrieved by ec ion requuiring clear text; and in_53 divided
re runs, 29 included equations using clear team, in 25 of which
dommats were retried by i ueh exertions.
18. Machi rational: Productive, idle and down time percentages
represennting thiarr~;eea t averages for the NINICARD equipment (cameras,
processor, sorter, duplicator and selector) were 36.11%, 42.6%, and 24.2%
respectively. Deatailed. breakdown for individual pieces of equipment was as
follows: Caamera 1: 24.6%, 56.7%, and 19.1%j Camera II: 32.7%, 118.3%,
annd l9.5?6j Processor: 37.6,, 1.5.8%, and 21.6%; licat?~r: 45.5%, 33.5%8
dad 28.1%j 9%, 35.1+%8 and 29.1%; Sew: 3 . 36.7'x, and 27.7+.
19. `3tartistics on documents in the co : Total documents amounted to
248633 of ch 063 were c . average size dcacument contained five
p,Wsp six f .le wqwwions and 16 code words. Totals and averages for numbers
of tagg,~ed, weards, images, documrtrts requiring more than one ISi IXCARD, cards used,
:First send seeconc So"eration cards produced, etc. were kept anus extrapolate.
'*r a full-scale MIMICARD o oration based on the following ass'.maptions:
a. 1000 documents and 20 requests mould be processed each day.
'?, pars:. 19: overall total includes Nodex material and reports Minicoded by
Mr Force.
*r4Preaeat daily inprut in Intellofax is cut 700 documents per dayy. Tie projection
to 7?040 was made to allow for anticipated increased volume and to assess a
e arability greater than that which we now have.
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
Wit?-l~F?I-I3~-]I~1~~~-Avg,
6
b. File expanded. Ninic s representing 70 documents would be
received from Air Force each day.
c. Minicezde aaaoimted is aperture cards and filed by document number
wul d. be used to service requests for specific documents c
do ai nt in use (or plannerd) for reproduction of the present
rture cards read be modified to reprice the liaicards
mowed in apertt cards*
a o Vital records would receive a sunmary listing of subject and area
codes with document n rs for each code and Mini garde mounted
in aperture cards.
mating cards with an electrostatic first l imsW of the
doh vauld be used as a source file.
S. Any now MENICAFD a ipawt procured woad have the owne operating
capabilities as that on hand*
h? Service perso?l would work a staggered shift and aback the
equipment before each day's operation.
Intesr9retiire ? ip v s ., r
21. test demonstrated clearly that, abject control of information and
the p s eilod thereto are the principal areas to consider in the develop-
mant of a successful system.
20 7he test revealed several vantages of A33IC&MM.) over Iutellofax. In
the main' these ware of a nature Stich could, with additional cost in manpower
or money, be incorporated into the Intellofaz w stem. The over-all retrieval ,
test ahoved, an appreciable qualitative advantage for C . This advantage
attributable chiefly to coding and procedural terabhiques developed by the
C-O-N-F- I-D -E-N-T -I-A-L
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
7-
Working Orwo during the test period. These would be largely transferable to
a revised Iaateellofax system.
23. The validity of con risoas berg ) NICAAD and Intellofax could
have been affected by the following factors; (a) the revised ISO and new Area
Code ware used. for ) CARD inprtt and the old ISC was used for Inctellrofaax; (b)
document coding was reviewed for ICNICAMD inp4at at the rate of one. In three
documents but such review occurred rarely for Intellofax; (c) dowmaent coding
transcript sheets were available for cross check men CARD retrieval van at
variance with Intellofax; (d) phrase coding, unlimited boundaries and logical
relaatiorthipa va3re available in 1ICARD but not Irrtellofax; (a) clear text
vas used in 1NF]I ICAIW only; Intellofaz used bibliographic citations. In the
111 11 of the Working Grroup, the revised ISC provided in V roved structure,
form and subject unity which was c for in part by Intellofax in this
test situation by normative procedures developed over a long period., and
familiarity with the application of the old ISO. The potential advantage of
MINTCARA input-review to provide more standardized and hi blxer quality input was
co ensaei ed. for by the fact that the Intellofa x team had had supervised training,
standard hater retation and published coding reminders in their p vio*as work
with the old ISO. Although dooanaaaauct coding sheets ware available only to
} MICA D for cases-reference check, the main file of the Intellofaax system
Nov idred. a source of control against the test dock of punched. cards. Phrase
coding, unlimited boundaries and logical xitlationships ware not available to
~Intellof . The extent to which this can, be added to present equipment is
limit+e+i. by the capacity of EAM for determing logical relationships by using
matching techaiiues and separate files for direct entry to major retrieval
paoblames. Clear text was an advantage to MIIIQ:CARD; clear teat oauld be used if
we re-designed the In tellofas card.
211. lion controlled by both I I ARD aan . Izxtelaofax incious:
subjects/coemaddties/organiaatiews/ 1 types; areaa/related area; modifiers
for subjects; security classification; locator umber; control nuanrber; publica-
tion date. Information controlled by MINICAR,') but not by Iutellofex includes:
names of persons, organizations or geographic locations; subject and commodity
specifications by clear text entry; modifiers based an Area Code (direction
indicators, nationality and moments codes); control of format (=Vs.. charts,
bibliegrap , ate.) by clear text entry. We have not fund informations-date
control to be of sufficient value to justify allocation of space in a re-
designed Inttellofaax system; the Gx*W re that the other categories be
studied for possible mcorporattion in a revised Intellofax systea:t, and a
p elite study has no v been. made. 1
25. The advantagey a s, then, of 1ICARD a re: clear text and phrass coding
capabilities and. file space savings.
26. MINICARD disadvantages nrt-tedd include the following:
a. Attempts to establish code-unit boundaries and, to provide linkage
between given DCIQICARD ids, within a phrase proved difficult for machine
iwOrocessinga To accamodate the ) ICABD voT;, the notation of the revised. ISO
Cm+;~-3hF-I-D. N-T-I?A?L
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
- 8 -
VMO
has been restricted to six digits; likewise natural language as identification
of a given item has been entered in artificial form as one MINICARD word. This
limited the value of natural language entries and necessitated rigid control
against lose of relationship or meaning within the MINICARD Record.
b. Indication of pagination within MINICARD was found impractical.
Each word within a given MINICARD phrase may be taken from a separate page of a
report processed. To paginate would require linkage and code-unit boundaries.
42. Although digital information in MINICARD can be supplemented,
deleted or altered by making a modified duplicate, this is a slow and laborious
process, amounting to the reprocessing of all information in the document. All
corrections must be photographic; item by item standarization via the gang-punch
technique is not possible. Conversion of a MINICARD file to reflect code changes
is difficult, and the use of systems change indicators, as suggested by Eastman
Kodak, is not the answer, according to our input people.
d. Mahal access to the detail file is a valuable asset in the
reference facility. MINICARD does not provide this facility.
e. Punch card equipment lends itself readily to dictionary building;
MINICARD equipment does wt.
f. The building of special purpose files as a by-product of the
detail Indexes is not feasible.
S. Identification of coding error is possible through mechanical
matching of detail entries against an established deck of approved codes. This
technique is possible with Intellofax but not with MINICARD.
27. Equi t: We have mentioned operating speed reductions engendered
by equipment moth cations. In addition, the block sorting operation (in which
the Minicarda are placed in the 100 magazine file blocks) which was to have been
performed mechanically, failed because of the d tolerances needed when
positioning the block over the transport belt.10 The overall assessment from
r:l
the equipment standpoint was that (a) it did not perform in accordance with the
specifications established at the outset of the project; (b) the modifications
of operating specifications would necessitate substantial maintenance and
supplies (and hence increase the requirement for standby equipment); (e) MINICARD
equipment does not easily lend itself to fit in as a subsystem of OCR's overall
machine system, much of which would have to be continued to process materials
not suitable for control by MINICARD.
Conclusions Con the on Pro act Domal
28. The expectation of economies in the Agency from a common comminity
program for one-time processing, coemron code, and identical equipment and
procedures hasn't come about and frem all indications would not come about
through MINICABD, whether this Agency adopted it or not. The Air Force is using
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
~C-0-1~F?I?A-I~?D~l'?I-A L ~i
.9-
MI1' CABD and does exchange information with us -? involving 70 documents per dad',
as compared with our present daily input of about 700 documents. Technological
advances since 1955 have been such that alternative systems have been presented
and no other agency in the USIB lens to use MINICARD. One of the aims of the
tJSIB Committee on Documentation (CODIB) has been the stimulation of compatibility
of systems considered, and this goal has by no means been reached; to hope for
identical systems is just not realistic. As to the common code, the ISC was
approv in principle as the TSIB community code and, hopefully, will be adopted
by all some day, but to date only Air Force and CIA use it. Moreover, the code
including the revision, was developed apart from MINICARD considerations and
applies regardless of the system.
29. As for I ICABD providing an inviolate storage file, the same is
provided by the aperture card/microfilm reel facilities of Inteliofax -? and ire
feel we must keep this manual access capability. Retrieval according to
association and other capability for more pertinent retrieval can be built into
Intellofax with the experience garnered from the test. As for speeds ,input is
slower for MINICAt because of its increa 3 complexity; this fact was noted
both during coding and preparation of machine logic. Access time did not prove
to be faster than Intellofax. it should be noted that the outlook is for iu pravm
ed EAM equipment: collator operating speeds of 1300 cpm vs present 480 cpm
selectors at 1000 cpm; sorters at 1000 cpm with further promise of doubling. By
the same token, second generation MItICABD equipment promises much improved
capability in speeds and other economies, such as the single machine combining
selector and sorter. Our test did not go into the relative merits of these
improved capeebilities.l4 Simultaneous processing was not really tested.
Retrieval quality (legibility) vas not impressive
30. The major econocty mentioned in the original project proposal was
space. Owing to the changed specifications, additional equipment hence additioaal
space would be required. There is no question but that the MINICABD files would
occupy less space than IBM cards; the ratio of Intellofax work and file space to
that of a two-card MINICARD system might approach 5:1 or mare. This is a
problem which must be tackled but it is note in our view, the paramount problem.
We assign higher priority to quality of input and relevance in retrieval. Steps
have been taken to reduce the need for hard copy document files and a proce lure
for further culling material from the active and inactive files in OCR and at
Records Center is under development.
31. The MtNICARD capability for greater depth of coding deserves mention.
The Working Group does not believe that depth of coding to the degree originally
considered is required -- such depth just isn't in the documents themselves. In
a small test-within-the-test sampling representing about one weeks production
(400 docents), 80% of the documents required three subjects or less; 91%
required five or less; 95.5% required eight or less. To the degree that this
sample is representative (and it is thought to be) if expensive equipment were
bought on the basis of providing greater coding depth, it would be applicable to
4-5% of the documents under control. Another wary of looking at the time expendi-
ture for input to )aNICARD is to consider that over 50% of the coding time was
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BOl 139A000200020010-9
VW#
F, r. a t on, leas then 7.0% of the d s :. /
Cone ndn m et 2Q9e cierecti .e
32. The revimA ISC provides for a precise .next theta, the 63A code
Dim the course of the test the Group was able to ideuti t' .301 subjects which
occurred with, a 'fioient fregiency that they should be (mind 'le) addeeL to the
i vision prior to its isouancek o Operational testing of the revi . code for
over a year afitor6c excellent experience for itr3 use with Ir l1efaa and d evel a l
a cadre of anal7aU trom moral OCR coc G nts o xetu,ed. to their
components with the advantage of this experience
33. No eigiificant conclusion were resche3 as to by-pxc u t information,
can, files management, nodazing po .icie ea utility of the he rt .e y collection*
retention value or purging, or the iasportance of the source card file. Clear
text was a MINICAFID plus and will be lied to Intel lofex . Mraw structure
r? not considered necessary by the Grou;p,, As to free vs. ' . wdifier
practice, the Group vould prefer the free modifier but the revised. ISO is bound.*
Coding practices and tools are being explored and. po lianin? r?, r?+eco a ions
d
ing
been made c :erning their a spllicati or to l.3tel, fate 3i: MAUS&& co
:off probab] y? be dwe by clerical per L o
31 Cenera y it was concluded that the ass .,& of subject control
developed during the team ate dl be used. in. redesigsa:lng Into lofa . The ~~+
diiegt}iy, . ionaLL~M1,6.1,.~? y, Try.asp/~Xnu~s~&, ~ce ~'forma and tran Script ahye a Pin's ya.~1.Li p~ icati:~n
for aids ~y
for A~u Division ses The d S'alue of .JS obIne listings as index ia aids
as demonstrated.
Kith Reference-to tht Over-all 0M*3 Flature
35- The Intellafaz, or aaay other ma him axe uteeae role 101 o a part of
dear service facility. Requewta for m chine searches a onSatitt to partial use of
the Library, difficult to pinpoint waurateltiy bWV estimate& a=t aba++xt 20%,?
vide cable use is also made of the Si, _c, Graphics, IL bu-trial and. Special
Registers. in the overall OCR picture, Intellofax ftarrdshes $i of the total
rafexjnces provided, and x e is 0. of the total reWEtE received (FY 59
figures) o With. increasi p seuree of bi dgst and. p o e`1 res ricticros,
ft.s., at present. It mar be that it can be nub .
In FT 59 them vare 2070 requests for Ia llofaz rum ?r3 313,277 Vista
levied on the office as a w iol e. Tecae w,s hi na :ru= f u=lacd. 809,7 581 selected.
referenies out of an office total of 1,0,1 ,335 references se .liM0 Other
rueste levied on OCR a lwat cert r resulted from refer ces faerniahed by
l 1l of runs but are not id 'tiflabl e as sum. * I : shor..d a "be noted that
the t,errs `'request" is an wabiguoua one w ich cased aim the Stoat from a telephonic
transaction completed, 'attile the c czar wits to a ubstantl at, rMAhie and. man-
hoar expenditure; no 'better unit of n aztarement than the t o*xr "request," has been
d ew as let,
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BOl 139A000200020010-9
 ILLEGIB
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
 Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
NOV
C-4-N-7-I-D-E-N-T-I-Ao-L
(2) identification of neared commodity iti ch is classed in a single code entry
having more than me cowmodity; (3) VxWaphic place nos (perhaps in the Bloc
only) . Provision for clear text entA7 of the following to be deferred for
further study: +s of research establishments; names of industrial firm;
identification of terms other than commodities; names of geographic locations
outside the Bloc; names of persons
e. I124 Card. Foxes: a proposed card design has a built-in
capacity for 32 additional types or iagpd Ilion. In a document recovery
systea,p variable types of information within a fixed field identified by tag are
as effective in retrieval as that derived from a one purpose field. This form
of information control can be readily adapted for input to a tic tape system.
and Decision
38. An a result of the test the Working Group concluded that CARD did
not live up to that had been hoped for in terms of our own problem. Miere were
demonstrable advantages, but ir4iortant disadvantages wu* also discovered. As a
consequence the Working Group did net recs ni a conversion from Intellofax to
MINICARL` : but recommended instead the msdification of Intellofax to incorporate
as many of the advantages of )4INICARD as ven?e technically and admin'strative:11,y
feasible. OCR monsagwomt took these findings into account., along with the
proportional role that ma^h4ne searches play in the overall, OCR service picture,
' and the present limitations on staff, money, and space. These considerations
led to a decision not to adapt the (CARD system as a. substitute for the OCR
Intellof .
Appendix: Source references and. selected bibliogx?aaphy on MINICARD.
C-OmN-FwIaD-B-N-T-I-A-L
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80B01139A000200020010-9
 25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9
Next 2 Page(s) In Document Denied
Iq
Declassified in Part - Sanitized Copy Approved for Release 2012/07/09: CIA-RDP80BO1139A000200020010-9