THE COMBINED FILE SEARCH SYSTEM
USER'S MANUAL
FOR
AN INFORMATION STORAGE
AND
RETRIEVAL SYSTEM
PREPARED FOR
CONSUMER PROTECTION AND
ENVIRONMENTAL HEALTH SERVICE
BY
THE SERVICE BUREAU CORPORATION
JULY 1968
-------
THE COMBINED FILE SEARCH SYSTEM
developed for
THE FOOD AND DRUG ADMINISTRATION
by
THE SERVICE BUREAU CORPORATION
USER'S MANUAL
PROPERTY OF
EPA LIBRAlff
RTP, NC
I~
-------
TABLE OF CONTENTS
INTRQDUCTION . . . . . . . . .
. . . . . . . .
. . . . . . . .
FILE DEFINITIONS
. . . . . . . . . . . . . . . . .
. . . .
DICTIONARY SYSTEM
. . . .
. . . . . . . . . . . . .
. . . . .
DICTIONARY LANGUAGE. . . . . . . . . . . . . . . . . . . .
DICTIONARY MASTER RECORD. . . . . . . . . . . . . . . . . .
TABBLD PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
DEDIT PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
DICT2 PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
DICTMSG PROGRAM. . . . . . . . . . . . . . . . . . . . . .
DPRINT3 PROGRAM. . . . . . . . . . . . . . . . . . . . . .
SYNTAPE PROGRAM. . . . . . . . . . . . . . . . . . . . . .
DPRINT4 PROGRAM. . . . . . . . . . . . . . . . . . . . . .
REL TERM PROGRAM. . . . . . . . . . . . . . . . . . . . . .
REL TRM1 PROGRAM. . . . . . . . . . . . . . . . . . . . . .
MAIN FILE SYSTEM.
. . . . . .
. . . . . . . . . . . . . . . .
FILE MAINTENANCE LANGUAGE. . . . . . . . . . . . . . . . .
MASTER FILE ORGANIZATION. . . . . . . . . . . . . . . . . .
MASTER FILE RECORD. . . . . . . . . . . . . . . . . . . . .
INVERTED FILE RECORD. . . . . . . . . . . . . . . . . . . .
CROSS REFERENCE FILE RECORD. . . . . . . . . . . . . . . .
MFEDIT PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
MFMAINT PROGRAM. . . . . . . . . . . . . . . . . . . . . .
MPRINT1 PROGRAM. . . . . . . . . . . . . . . . . . . . . .
IDUPDATE PROGRAM. . . . . . . . . . . . . . . . . . . . . .
UPDXREF PROGRAM. . . . . . . . . . . . . . . . . . . . . .
XREFPRNT PROGRAM. . . . . . . . . . . . . . . . . . . . . .
I VUPDT PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
I FPRNT PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
IVPRINT PROGRAM. . . . . . . . . . . . . . . . . . . . . .
DUMPCELL PROGRAM. . . . . . . . . . . . . . . . . . . . . .
LOADCELL PROGRAM. . . . . . . . . . . . . . . . . . . . . .
SEARCH SYSTEM
. . . . . . . .
. . . . . .
. . . . . . . . . .
SEARCH CONTROL PARAMETERS. . . . . . . . . . . . . . . . .
SEARCH LANGUAGE. . . . . . . . . . . . . . . . . . . . . .
SEARCH PROGRAM. . . . . . . . . . . . . . . . . . . . . . .
Ie
2
6
9
19
22
23
28
34
36
39
43
44
50
52
54
64
72
77
79
80
81
83
86
89
91
93
95
97
99
102
104
105
108
114
125
-------
INTRODUCTION
The information storage and retrieval system;developed,by The Food and Drug
Administration primarily-for storing and retrieving information related to
adverse drug reactions. is also applicable to many other disciplines. Vir-
tually any document whose contents can be accurately described by words or
symbols (descriptors or index terms) and numeric data can be stored and re-
trieved by the system. In addition to storing the descriptive data contained
in the searchable portion of a document, narrative text may be stored and re-
called when that document satisfies the criteria specified in a search request.
The system is designed for manual indexing where an indexer describes the sig-
nifi~ant portions of the contents of a document through the use of Descriptors
or Keywords. <- To___refine the meaning of the descriptor, we have introduced the
qualifier. A qualifier may consist of a subdescriptor, a numeric value, or a
combination of both.
Since the system is designed to operate in a direct access environment, care
must be taken not to bulk the files unnecessarily. For this reason, several
types of descriptors are defined in the system. They are: PRECISE, COMMON,
TEMPORARY, and Ir.
PRECISE:
A Precise descriptor is a keyword which generally doesn't occur is a high
percentage of the documents in the system and hence distinguishes a document
as different from most other documents. A useful way to determine if a de-
scriptor should be designated as precise is to consider how many documents
would be retrieved if a query was to be made based only upon that particular
descriptor. If the number would be very excessive, the term should probably.
although not necessarily, be Common.
COMMON:
In contrast to the precise descriptor, a Common descriptor is one that is con-
tained in too many documents to be very discriminating for searching purposes.
For example, asking for all MALES in a medical data base would result in re-
trieving about one-half of the file. That very question can be handled by
the system, but not by the classical inverted file approach.
-------
TEMPORARY:
When new terminology is added to the system's vocabulary, it may not be
determined whether a term should be designated as precise or common since
not enough is known about the frequency with which the term will be used
as a descriptor. In this case, it would be assigned a temporary precise
function and could be reclassified later.
pre-
ID. :
All descriptors validly indexed into documents resident in the system are con-
tained in the searchable portion of the document file records. In addition,
common descriptors are extracted for inclusion into the identification (10.)
file records. All records in the document file containing any common descrip-
tors will produce a corresponding entry in the 10. file. The 10. file is
searched when a query is made based solely upon common descriptors. In ad-
dition to the common descriptors, the user can specify other descriptors to
be included in the 10. file by classifying them as 10. file descriptors.
A number of keyword oriented systems fail to carry the intelligence related
to the context in which a descriptor has been used, For instance if a
record contains the descriptor RASH, how does the system tell whether RASH
is a symptom for which a drug was administered or a reaction caused by a drug.
This system makes such a distinction possible through the introduction of
file/context codes which may be associated with a descriptor either by the
indexer or automatically by the dictionary.
In a well implemented information storage and retrieval system, care must be
exercised to assure proper use of the indexer's vocabulary, Once an improperly
indexed document enters the system, it may be extremely hard to find. Ob-
viously, the system doesn't detect the improper use of a term from a conceptual
point of view, but it does diagnose misspelling, use of a term as a descriptor
when it is not allowable as such, indexing a document with a descriptor which
is not valid for that particular data file, and use of a non-preferred term
(in which case the preferred term is automatically substituted),
Vocabulary control is accomplished through the use of a system dictionary,
The dictionary contains the vocabulary of index terms (descriptors) which the
indexer may use in describing documents, Each index term is entered in the
dictionary in ascending alphabetic sequence, Subordinate to each term are
certain related data such as:
1.
List of dictionaries to which the term belongs
-3-
-------
2.
List of file/context codes
3.
Preferred or USE term (if a synomym exists)
4.
Related terms (if any occur in the dictionary)
s.
See also terms (as desired)
6.
Scope notes (as desired)
Related terms are treated in a special way in this system. Any related terms
stored under a main term are automatically added to a document whose contents
are described by that main term. This amounts to a form of semi-automatic
indexing.
-4-
-------
INFORMATION SYSTEM
M ASTE R
FILE
DES C R I PT 0 R
INDEX
UPDATE
IN-
VERTED
FILE
-MASTER FILE
TERM
TERM
QUALIFIER
TER M
-INVERTED FILE
TE R M:# # :#
--
TERM # :# :#
---
EDIT
FORMAT
VALIDATE
MASTER
DOCUMENT
FI LE UPDATE
CROSS
REFERENCE
UPDATE
XREF
FILE
-5-
FDA
CONTROL
DICTION
ARY
I. D.
FI LE
UPDATE
-CROSS REFERENCE FILE
EXT.!!..- I NT ~
EXT~INT ~
-I,D.F/LE
COMMON
COMMON
COM M ON
-------
The System maintains five files which are basic to the system.
They are:
1.
Dictionary File
The dictionary is primarily designed to control the vocabulary used by
the indexer and to serve as a guide to the searcher in formulating re-
quests. An important feature of the dictionary maintenance subsystem
is the ability to produce a variety of listings which may be used by
personnel maintaining the vocabulary and can be distributed to the
users of the system. The listings may also be a useful aid in standard-
izing a vocabulary related to a particular discipline such as the EJC
thesaurus developed by the Engineering Joint Council. ..
2.
Document File
The document file is the heart of the retrieval system since it contains
the documents stored in the system. All descriptors used in indexing
documents (whether they are precise, common, or temporary), their con-
text codes, their subdescriptors and numeric values, and their associated
links are stored in the searchable portion of the document file records.
Textual information realted to a document is stored in "free text" seg-
ments which immediately follow the searchable segment.
3.
Identification File (10.
File)
The ID. file records are subsets of the information contained in the
searchable segment records of the document file. All common descriptors
and their qualifying data are contained in the 1D. file for any given
document along with other descriptors which have been specifically tag-
ged for entry into the 10. file.
4.
Inverted File
The inverted file is organized in ascending alphabetic sequence by de-
scriptor followed by a list of all documents in which the descriptor was
used as an index term. If a descriptor was used in 500 documents, a
list of 500 document numbers would be appended to the descriptor. Al-
though common descriptors appear in the inverted file, a list of their
document numbers would not be included. The inverted file serves as a
high speed screening aid to be used by the search programs.
-6-
-------
5.
Cross Reference File
When a document is indexed into the system, it is assigned an "internal
document number" which will differ from the manually assigned document
number in almost all cases. Only coincidence would cause them to be
identical. In order to allow the searcher to query based upon an 'ex-
ternal" or manually assigned document number correspondence must be
maintained between the manually assigned document number and the number
assigned by the file maintenance system. Also, duplicates being added
to the system are rejected by means of determining whether the document
already exists within the file. This option may be overridden by the
indexer if duplication of external numbers is desirable.
The system allows several types of searches to be performed:
1.
Document Search
In the document search, the searcher must enter a list of the document
he wishes to retrieve. He may request the documents by either internal
or external numbers. If the request is by external number, the first
number in the list must be preceded by an 'X'.
2.
Mixed Search
In the event the searcher may wish to retrieve a predetermined list of
documents which satisfy some logical criteria, he enters the document
numbers followed by a Boolean logic statement representing the logical
constraints which must be satisfied by the contents of the documents.
3.
Boolean Search
The Boolean search is used when the searcher wants to retrieve all docu-
ments in the system which satisfy a specified list of criteria. In this
case, the searcher specifies the list of criteria necessary to be satis-
fied for a document to be retrieved in the form of a Boolean logic
statement.
-7-
-------
In designing a generalized information retrieval system, one cannot conceive
of all of the reports a user may wish to derive from the system. For this
reason the system has a variety of standard printouts which may be obtained.
In addition to the standard printouts, the searcher may request a report tape
which will contain selected data in a fixed format which he may process by
use of his own specialized programs.
-8-
-------
THE DICTIONARY SYSTEM
- 9-
-------
DICTIONARY SYSTEM
This series of programs, known collectively
designed to produce tape, printed, and disk
ary terms comprising the vocabulary used in
retrieval system.
as the dictionary system, is
file versions of the diction-
the information storage and
The dictionary serves two major functions within the total system. First
it is used as an automatic validation device against which both new docu-
ments entering the system and requests for information from the system are
checked. This validation ensures that only those terms which have been
approved as acceptable descriptive terms are used to describe the docu-
ments, and likewise, that inquiries to the file are phrased in acceptable
terms. The second major function of the dictionary is to provide printed
copies of the acceptable terms for use by document indexers, requestors,
and other interested parties. Terms may be added to, deleted from, or
changed within the dictionary and new versions prepared (in either or both
machine readable and hard copy form) as desired.
The system consists of a series of nine programs which perform the various
functions required. They are:
1. TABBLD (Table Builder)
2. DEDIT (Dictionary Edit)
3. DICT2 (Dictionary Maintenance)
4. DPRINT3 (Dictionary Print)
5. DPRINT4 (Dictionary Print)
6. SYNTAPE (Synonym Tape)
7. DICTMSG (Dictionary Error Message Print)
8. RELTERM (Subordinate Term Extractor)
9. RELTRMl (Subordinate Term Validator)
10-
-------
DICTIONA RY SYSTEM
GENERAL DATA FLOW
TERM
MAINTENANCE
SY S RDR
DICTIONARY
EDIT
PROGRAM
DSORT
DSORT
DICTIONARY SYS
MAINTENANCE 010
PROGRAM SYS
~II
CONTROL
CARD
CONTROL
CARD
SY NTAPE
PROGRAM
DPRINT3
PROGRAM
DICTMSG
PROGRAM
PUBLISHED OR
WORKING
COpy
SELECTE D
DICTIONARY
LIST
I) ERRORS
2) DELE TI ONS
3) ADDITIONS
4)CHANGES
-11-
-------
The basic operations of the system are handled by two programs: DEDIT
(Dictionary Edit) and DICT2 (Dictionary Update). DEDIT accepts the
punched cards (or tape equivalent) representing the data to be placed
in the dictionary. It validates the data, ensuring that various con-
ditions are met, and converts it into a form suitable for further pro.
cessing. Errors are printed as encountered in the data.
12-
-------
DICTIONARY MAINTENA NC E
CARD
INPUT
LISTING OF
INPUT DATA
D ED IT
DSORT
DICT2
-13-
DICTIONARY
ON DIRECT
ACCESS
DEVICE
CARD
IN PUT
TAB BL 0
CATALOGABLE
TABLES
TABLES IN
CORE
IMAGE
LI BRA RY
-------
After the edited information has been sorted in sequence using a standard
sort program, it is ready for processing by DICT2. This program accepts
the edited input tape and the previous dictionary master tape (if any).
The output from DICT2 consists of a new master dictionary tape, a new
disk file version of the dictionary, Bnd a tape of d1ctionary changes
which resulted from the combining of the new input with the previous dic-
tionary file. This change tape also contains error messages for invalid
data, if any. A schematic illustration of the functjons of DEDIT and
D1CT2 is shown below.
One of the functions of the DEDIT program is to convert the three letter
mnemonic codes used to identify the various dictionaries within FDA
(e.g., HUMan, VETerinary, etc.) into numbers, since these codes are
stored as numbers wi thin the files of the system. For tilis purpose,
the DEDIT program requires access to a table of equivalents which gives
the number corresponding to a given three letter code. The table can be
created, modified. o~ added to at any time by use of the TABBLU llaDle
Buildin~gram. This program accepLs up LO L~o
-------
DICTIONARY PRINTI NG
CONTROL
CARDS
DPRINT3
SYNTAPE
SELECTIVE
D ICTI ONARY
LISTING
DSORT
MAINTENANCE LISTING
PUBLISH OR
WORKING
COPY
LIST OF
MISSING
SYNONYMS
DI CTMSG
LISTING OF
MAINTENANCE
PER FOM ED
-15-
-------
The program DICTMSG is used to print the tape of changes, deletions, ad-
ditions, and errors resulting from execution of DICT2. Certain options
are provided in DICTMSG to permit printing of only certain classes of
data (e.g., print only the additions, or only the errors, if desired).
DICTMSG is intended only as a maintenance tool, to provide printed in-
dications of what changes have occurred. It is not intended to produce
"pub lishab Ie" copy.
DPRINT3 accepts the dictionary tape as input and produces a printed copy
of the dictionary. This program, unlike DICTMSG, operates on the full
dictionary tape, and can print all dictionary entries. It is provided
with a number of options to permit selective printing of only certain
entries, or of only selected portions of data from entries. It can,
for example, print only the SEE ALSO data from selected entries.
DPRINT4 is the program which produces publishable copy from the master
dictionary tape. It produces output in two formats: 8 1/2" x 11" pages
and 14 x 22 inch pages. The latter is designed for photo-reduction and
subsequent publication. DPRINT4 can be used in two different modes. In
the first mode, the dictionary tape is simply printed as it appears on
tape. In the second made, "reverse cross references" are included. For
example, the dictionary may contain the following three entries:
ACETYLSALICYLIC ACID
HEll CON
use ACETYLSALICYLIC ACID
MEASURIN
use ACETYLSALICYLIC ACID
In the first mode, those three entries will appear as shown above. How-
ever. in the second mode, the ACETYLSALICYLIC ACID entry will appear as
follows:
ACETYLSALICYLIC ACID
used for
HELICON
used for
MEASURIN
-16-
-------
DICTIONARY VALIDATION
RELTERM
DSORT
RELTRMI
LI STI NG OF
TERMS MISS-
ING AS MAIN
TERMS
-17-
CONTROL
CARD
I) Related Terms
2) Use Terms
3) See Also Terms
-------
This has the effect of bringing together under the preferred term
(Acetylsalicylic Acid in this case) all the other terms which refer
to it. In order to locate and sequence these "see references" so that
they will appear under their appropriate preferred term, antoher pro-
gram, SYNTAPE (synonym tape generator) is used, followed by a standard
sort. The sort produces a tape which is then merged with the master
dictionary tape, producing a combined listing of both the dictionary
entries themselves and their synonyms. .
The subordinate term validation programs ensure that each term used
in the dictionary as a related term, a use term, or a see also term
occurs as a main term as well. This maintains integrity within the
dictionary, eliminating blind references because of misspelling or
revision of terms. RELTERM extracts the subordinate terms and RELTRMI
matches each with a main term. Any terms not found as main terms are
listed with the main term under which they occur.
-18-
-------
DICTIONARY LANGUAGE
The Dictionary Language is designed for the use of the editorial staff
charged with the maintenance of the FDA Dictionary System. It offers
a means of adding new terms to the dictionary, changing the various
elements comprising an entry, and of deleting which are no longer re-
quired terms. . .
The dictionary consists of a variable number of "entry terms", each of
which is followed by some descriptive information, known as the "term
data". An illustrative sample of an entry term followed by term data
is given below.
ASBOCI LLIN
FUNCTION-- PD,SD DICTIONARIES--115
FILE / CATEGORY NUMBERS--
163-157,1-14
RELATED TERMS--
ANTI B lOTI CS
P PENICI LLINS
SCOPE NOTES--
MAJOR INDEX TERM
AUTH-M
C PENICILLIN G POTASSIUM-C
PROCAINE PENICILLIN
ABBOCILLIN
In this illustration, ABBOCILLIN is the entry term. The information
following it is the term data. In this particular case, the following
elements are present.
FUNCTION--PD,SD
Indicates that this term is valid for
use as a precise descriptor or a sub-
descriptor
DICTIONARIES--115
Indicates that this term is a member
of the FDA Dictionary identified as
number 115
FILE/CATEGORY NUMBERS--
153-157,1-14
Indicates that this term has been as-
signed the File/Category (context) codes
shown
RELATED TERMS--
ANTIBIOTICS--etc.
These terms have been assigned to ABBO-
CILLIN because they are closely related
in their meanings and usage within FDA.
Every time ABBOCILLIN is used to describe
a document indexed into the master file
these related terms will automatically'
be assigned to that same document
-19-
-------
SCOPE NOTES--
Variable contents, as assigned by the
individual who entered this term in the
file
MAJOR INDEX TERM
AUTH-M
ABBOCILLIN
Another type of term data is permitted as well: SEE ALSO references. They
are similar in form and content to RELATED TERMS, but are included only
for the indexer's benefit in printed versions of the dictionary. They are
not automatically included in the indexing of any given document. This
particular term has no SEE ALSO references.
The dictionary editor may use any or all of the above elements to describe
a dictionary term. The only one which is mandatory is FUNCTION. The
editor must indicate whether a term is a precise, a common, or a temporary
term so that the term will be handled in the proper manner when incorpor-
ated into the indexing of some document.
In addition to entries of the nature shown above, a second kind is per-
mitted, the 'USE' reference. An example of such an entry is shown below.
NAPTHYL SALICYLATE
FUNCTI ON-- PD
DICTIONARIES--ALL
USE--
NAPTHOL SALICYLATE
The entry for the term referred to is shown below.
NAPTHOL SALICYLATE
FUNCTION-- PD
FILE / CATEGORY
1-66,163-179
DICTIONARIES--115,2
NUMBERS--
RELATED TERMS--
ANTISEPTICS
SALICYLATES
USED FOR--
NAPTHYL SALICYLATE
SCOPE NOTES--
MAJOR INDEX TERM- A-NAPTHYL
AUTH-M
SALI CYLATE
-20-
-------
If NAPTHYL SALICYLATE is ever used by an indexer to describe some docu-
ment being added to the FDA file, the dictionary programs will auto-
matically substitute NAPTHOL SALICYLATE.
The 'USED FOR' entry under NAPTHOL SALICYLATE is of particular interest.
These USED FOR entries are never established by the dictionary editorial
staff. They are an automatic result of the execution of certain of the
dictionary programs (SYNTAPE and DPRINT4). Every 'USE' reference is
automatically inserted in its appropriate place under the term referred
to, so that all synonymous terms are brought together for inspection
under the preferred term. ..
If an editor should inadvertently construct a 'USE' reference referring
to a term which is not in the dictionary, the programs will detect this
error and print a message to that effect. Every term referred to in a
'USE' entry must be an entry in the dictionary. It must, of course, not
be another 'USE' entry.
In preparing input for the dictionary maintenance system, the editors
will use a "free-form" card format, in which various elements of the
information may be entered in a reasonably unconstrained manner. Care-
ful attention must be paid to the punctuation, however, because these
punctuation marks are used to distinguish the information itself from
the "control data" which indicates the kind of information being dealt
with. The details of these input formats are contained in the description
of the Dictionary Input Edit Program. The internal format of the dictionary
record in the system is shown below.
-21-
-------
DICTIONARY RECORD
LENGTH
CONTENTS
2 bytes
2 bytes
20 bytes
20 bytes
2 bytes
2 bytes
1 byte
1 byte
1 byte
0-510 bytes
1 byte
1 byte
1 byte
1
-------
TABLE BUILDING PROGRAM (TABBLD)
Purpose
TABBLD is used to build tables of dictionary, file, and context mnemonics.
After determining the validity of the input data, it sorts that data and
then punches the complete job stack deck to permit loading and cataloging
of the tables required for the dictionary input edit program (DEDIT).
Input
1.
Requirements
A control card must corne first, followed by from 1 to 256 input cards.
The dictionary table may have only 255 entries.
Format
2.
a.
Control Card
For each run there must be one and only one control card. There
are three permissible formats, all of which begin in column 1.
They are:
1. DCT~:
2. FI L~:
3. CA T~:NNN where NNN represents a file name. Please
refer to the description of input cards
for a more detailed specification.
b. Input Cards
Input cards have the following format:
1.
NNN XXX
where NNN represents the file name
mnemonic. The first character which
appears in column 1 must be alphabetic.
The other characters may be alphabetic
or numeric with trailing blanks. No
special characters or embedded blanks
are allowed.
-23-
-------
Column 4 is left blank.
XXX which is punched in columns 5 through 7 symbol-
izes any number from 000 to 255. Each number must
be used only once in each data set, but the num-
bers need not be in sequence. For the dictionary
table, the number may not be zero, since zero is
used to indicate that a term is included in all
dictionaries.
The remaining columns will normally be left blank,
but they may be used for comments. The program
does not check them in any way but they will ap-
pear on the printed listing of input.
Processing
The program reads the first card which must be the program control card and
set up the 'phase' card in the output format area. It tests whether the
card is a DCT~:, or CAT~: card and, if it is a CAT~: card, checks that
the filename mnemonic (columns 5 through 7) is valid. If no valid control
card is found, the program prints an error message and cancels the job~
When a valid control card has been found, the program reads and prints the
input cards. A running count is kept to check that no more than 256 cards
are entered. As each card is read, it is checked to be certain that the
number in columns 5 through 7 is between 000 and 255 and is unique and in
the current data set. The filename mnemonic in columns I through 3 is
checked for validity. Whenever an error is encountered, the program sets
a switch and prints a diagnostic message. If no errors have been found in
the set, the first seven card columns are stored. The program continues
to read, print, and test the input cards until the end-of-file is reached.
After all the input cards have been
whether any errors were logged. If
and the job is cancelled.
read and processed, the program tests
there were errors, a message is printed
If no input errors were detected, the program performs an internal sort
on columns I through 3 of the input. If identical mnemonics are found,
a switch is set.
After the sorting has been completed, the program tests whether any
mnemonics were repeated, and if so, determines which ones were dupli-
cated, prints an error message, and cancels the job.
-24-
-------
If there was no duplication, the program tests whether 256 cards were
entered and, if not, arranges to 'pad' the punched output. The dic-
tionary input edit program requires that the tables have 256 entries.
'Padding' is simply the filling of sufficient extra positions to bring
the total to 256.
It then sets up and punches the complete deck required for loading and
cataloging of the tables used in the dictionary input edit program.
Output
1.
Card Formats and Explanatory Notes
a.
Card Formats
1.
II JOB CATALOG
2.
II OPTION CATAL,LINK
3.
PHASE DCTT AB, ~~
PHASE FI L TAB, ~~
PHASE CATNNN, ~~
The characters in columns 10 through 12
correspond to columns 1 through 3 of the
input control card. NNN represents the
filename mnemonic characters which ap-
pear in columns 5 through 7 of a CAT~~
type input card.
4.
II EXEC ASSEMBLY
5.
CSECT
6.
DC C'XXX'
DC FLl' XXX'
The XXX's in this pair of, cards repre-
sent the filename mnemonic and corres-
ponding number. The number of pairs of
DC cards will be equal to the number of
input cards.
7.
DC NNNX'FFFFFFFF'
A filler card is punched only if there
were fewer than 256 input cards. NNN
equals the difference between 256 and
the number of input cards.
8. I~~
9, II EXEC LNKEDT
10. 1&
-25-
-------
2.
Listing of Sample Output
For the protection of the user, this program produces punched output
only when all of the input was valid. A listing of sample output is
given on the following page,
-26-
-------
I
~.
~
~
\
Listings of Sample Output
1.
Valid Input
CAT~~VET
PHB 002
PHA 001
CHC 004
ING 003
MIT 006
DRG 005
THIS SPACE MAY BE USED FOR COMMENTS
COMMENTS WILL NOT BE CHECKED IN ANY WAY BY THE PROGRAM
2.
Invalid Input
PHA .0.01
HB .0.02
IT IS INVALID TO USE SPECIAL CHARACTERS AND LEADING OR EMBEDDED BLANKS IN COLUMNS 1-3.
ONLY ALPHABETIC OR NUMERIC CHARACTERS AND TRAILING BLANKS ARE ALLOWED.
MIT 299
THE NUMBER IN COLUMNS 5-7 MUST FALL BETWEEN .0.0.0 AND 255.
C.02 .0.04
DRG .0.01
ANY GIVEN NUMBER
REPEATED.
IN COLUMNS 5-7 MAY BE USED ONLY QNCE IN EACH DATA SET.
THE NUMBER 001 HAS BEEN
AB/
THE NUMBER IN COLUMNS 5-7 MUST FALL BETWEEN .0.0.0 AND 255.
-27-
-------
DICTIONARY EDIT (DEDIT)
Purpose
The function of DEDIT is to examine the input information submitted on card
or tape records. When the data and operations involved have been evaluated,
the information is rewritten in a standard format, If all information for a
term is acceptable, the records formed for it are written on tape. Other-
wise, errors which have occurred are noted on the printed output,
DEDIT permits two broad classes of changes to be made to the dictionary
file, Individual dictionary entries can be added, deleted, or modified
as desired, In addition, words or phrases used as related terms, see
also references, and the like can be revised or deleted wherever they ap-
pear through use of the 'block change' feature,
Input
Input to this program is punched into cards according to the formats de-
scribed below, General constraints which apply to DEDIT input are:
1.
A dictionary term may consist of not more than 64 alpha-numeric char-
acters. Seven special characters are also permissible: right and left
parentheses, hyphen, slash, comma, period, and pound sign,
2,
All input cards must be in ascending sequence based upon digits punched
in columns 1 through 4, These numbers need not be consecutive,
3.
All block change cards must precede any changes to individual diction-
ary records. Block change cards may be numbered independently from
individual change records (i,e., block change cards may be numbered
from 1 to n, immediately followed by individual term changes numbered
from 1 to n),
4,
The first input record must be a BCH (block change header) record or
a HDR (dictionary term header) record,
5.
All ADD, DELETE, and REPLACE records may include continuation
These continuation cards contain a sequence number in columns
through 4, with columns 5 through 10 blank. The continuation
begins in column 11.
cards.
1
data
-28-
-------
If there is a character punched in column 80 of a card and a character
punched in column 11 of a continuation card, it is assumed that there
is no blank between the two characters.
6.
All SCOPEnn (scope note) cards must be
a given dictionary term. Only another
same term, a HDR card - indicating the
term--or an EOF (end-of-file) card may
7.
Block Change Cards
the last cards pertaining to
SCOPE card, relating to the
beginning of a new dictionary
follow a SCOPE card.
1.
Block change header must be the first of the block change cards; only
one is accepted. The format is:
Columns
Columns
Columns
1 4
5 7
8 - 80
Card Sequence Number
'BCH'
Blank
2.
Block change trailer must be the last of the block change cards; only
one is accepted. The format is:
Columns 1 - 4
Columns 5 - 7
Columns 8 80
3. Delete
Columns 1 - 4
Columns 5 - 10
Column 11
Columns 12 - 80
4.
Replace
Columns
Columns
Column
Columns 13 -
1 4
5 - 11
12
80
Card Sequence Number
'BCT'
Blank
Card Sequence Number
'DELETE'
Blank
Term name followed
any combination of
semicolons.
by a colon followed by 'ALL' or
'SA', 'RT', 'USE' separated by
Card Sequence
'REPLACE'
Blank
Term name to
lowed by the
Number
be replaced followed by a colon fol-
replacing name.
An example of two different block changes is given below:
0001 BCH
0002DELETE ACUTE POLIOMYELITIS: RTjSA
0003REPLACE ACUTE POLIOMYELITIS: POLIOMYELITIS
0004BCT
-29-
-------
These cards will result in the deletion of the term ACUTE POLIOMYELITIS
wherever it appears in the dictionary file as either a related term (RT)
or as a see also (SA) entry. If the term ACUTE POLIOMYELITIS appears
other than as a RT or SA entry in any dictionary record, it will be re-
placed by the single work POLIOMYELITIS, as directed by card ~~03 above.
Since the DELETE card above eliminated the term from all RT and SA oc-
currences, the only possible instance of replacement would be in the USE
entries.
Note that the block change DELETE does not result in the elimination of
the term as a main dictionary entry. The term is deleted only where it
is embedded within the data under a main entry. To delete a main dictionary
term it is necessary to use the dictionary maintenance cards described
below. A main term cannot be deleted from the dictionary by use of the
block change.
Dictionary Maintenance Cards
1. HDR card
Columns 1 - 4
Columns 5 - 7
Column 8
Columns 9 - 80
2.
ADD cards
Columns
Columns
1 -
5 -
Column
Columns
8
9 - 80
New Term
Designation
Function
Card Sequence Number
'HDR' (designates the beginning of a set of data
pertinent to a term)
Blank
Blank if a new index term is being entered into
the dictionary. Term name if maintaining exist-
ing dictionary term.
4
7
Card Sequence Number
'ADD' (indicates that a new term is being added
to the dictionary or that information is being
added to an existing term)
Blank
As described below:
'NT: Term Name;'
Must be present if there was no name on the 'HDR'
card. Must be the first of the 'ADD' card entries.
Only one NT: is allowed per 'ADD' data set.
PO
, F: (TO; SO; 10) ; ,
CD
If the term may be used as a descriptor in the
master file, either PO, CD, or TO (never more
than one) should be entered. A function must
-30-
-------
Dictionary
File/Context
See Reference
Related Term
See Also
3.
Delete Cards
Columns
Columns
1 - 4
5 - 10
Column
Columns 12
11
80
be given whenever NT: is present. Only one F:
entry is allowed per data s6t, but may contain
than one entry (e.g., F : (DD;SD;ID)).
xXPD - Precise Descriptor
xXTD - Trial Descriptor
~mCD - Common Descriptor
~'~'SD - Subdescriptor
xXID - Identification File Entry
( )
more
'DICT: (DCT;DCT;...);'
This entry specifies the dictionaries to which the
term belongs. Up to 20 dictionaries may be speci-
fied for a term. Only one DICT: entry per 'ADD'
data is Dermitte~ Eac~ DCT is a 1-3 character entry
" ,
'FC: FIL-CXT;'
This entry specifies the files in which a term
may appear and the context in which it is used
within that file. Up to 20 FC: entries may be
specified for each main term or related term.
Each FIL and CXT is a 1-3 character entry.
'USE: Term Name;'
This entry specifies a preferred term which is
to be substituted for the index term following
the NT: entry. Only one USE: may be entered
for a term.
'RT: Term Name;'
Up to 40 related terms
dex term. None may be
specified.
may be entered for one in-
entered if 'USE' has been
'SA: Term Name; ,
Up to 40 see also
index term. None
was specified.
terms may be entered for one
may be entered if a 'USE' term
Card Sequence Number
'DELETE' (indicates that
follows is to be deleted
tionary record)
Blank
As described below (DELETE card entries have the
same restrictions as do the ADD card entries ex-
cept that the FC: entries may only follow the
main term) : If the file/context entries following
a related term must be changed, the related term
must be deleted and re-entered with the correct
file/context data.
the information which
from an existing dic-
-31-
-------
Function
Dictionary
See Reference
Related Term
See Also
File Context
Scope
4. Replace cards
Columns 1 - 4
Columns 5 - 11
Column 12
Columns 13 - 80
See Reference
Related Term
See Also
5. Scope cards
Columns 1 - 4
Columns 5 - 9
Columns 10 - 12
Column 13
Columns 14 - 73
Columns 74 - 80
PO
'F: (CD;SD;ID):'
TO
'DICT: (DCT;DCT;...);'
'USE: Term Name;'
'RT: Term Name;'
'SA: Term Name;'
'FC: FIL-CXT;'
'SCOPE: ALL or Nl;N2;...Nm'
where N is a 1-3 digit number from ~ - 255.
Up to 256 note numbers may be specified. If
SCOPE is specified, it must be the last entry
in the DELETE data set.
Card Sequence
'REPLACE'
Blank
As described
Number
below:
'USE" Term to be replaced; Replacing Term;'
'RT: Term to be replaced; Replacing Term;'
'SA: Term to be replaced; Replacing Term;'
Card Sequence Number
'SCOPE'
Scope note line number
Must be a three digit number ~~l - 255 with any
necessary leading aeros.
Blank
Scope Note (all characters are valid)
Blank
An example showing the addition of a new term to the dictionary and the
revision of an existing one is given below.
-32-
-------
0001
0002
0003
0006
0007
0008
HDR
ADD NT: EMPIRIN; F: (PD); DICT:(DRG);
FC: HUM-DRG; RT:ACETYLSALICYLIC ACID
SCOPE040 REF/AMERICAN DRUG INDEX
HDR MEASURIN
ADD DICT: (DRG); RT:NON-NARCOTIC
Card 0001 is a header card. It is blank from column 8 because it precedes
the addition of a new main term to the dictionary. The term itself is
EMPIRIN, shown following the code NT: in the first ADD card after the HDR,
Card 0003 is a continuation of the data about EMPIRIN, giving the File/
Context code and a related term. A single line of scope note is given for
this term as well. Note that the line number need not be 001. It may be
any number desired. However, the scope note will be assigned a different
nu~ber internally so that additional scope notes may be inserted between
lines, if desired.
Processing
DEDIT is the buffer between the people who create update data for the master
dictionary file and the program (DICT2) which updates the file. As such, it
edits and reformats raw data and flags certain erroneous data with messages
explaining the nature of the error. The format of the input data is ex-
plained in the INPUT section. DEDIT checks each card to make sure that it
conforms to the basic input format. If it does, the nature of the modifi-
cation requested is determined. If the request is valid, records are
created for input to the dictionary maintenance program. All cards per-
taining to that request are processed. When a new request is encountered,
the records thus created are output on tape, Then the new request is
processed, Because of the order in which modification data must be sub-
mitted to DEDIT, all block changes will be processed first and will be
written at the beginning of the output tape, Modifications to individual
dictionary entries will follow. When an end-of-file occurs on the input
file, processing will be terminated,
Output
DEDIT produces a tape of changes on 5Y5005. These records, once sorted,
provide the input for DICT2. In addition to the tape of valid changes,
error messages pertaining to faulty input data are printed along with the
data to which they pertain.
-33-
-------
DICTIONARY UPDATE PROGRAM (DICT2)
Purpose
The purpose of the dictionary update program is to update the dictionary
update file (tape). This program also builds the dictionary file disk.
Input
1.
Mount the sorted update tape on SYS~~6. This tape is output from
the dictionary edit program and then sorted on main term.
2.
Mount the prior dictionary file tape on SYS~~7.
Processing
After opening the various files and initializing switches and counters,
this program reads up to 20 block change records from the sorted update
tape and stores them in core. Each dictionary record is read from the
prior dictionary tape. If the update records compare low or equal to
the dictionary record then some sort of updating is done (block change,
add, delete, or revise) unless there is an error, such as an attempt to
delete a non-existent dictionary term. For each add, delete, or change,
three or more records are written on the message tape. For example, if
a scope note has been added, then four records are written on the mes-
sage tape; a change message, a message stating that a scope note has
been added, the prior dictionary record and the new dictionary record.
If there is an error, one or more error messages will be output indicat-
ing the type of error.
All unaffected dictionary records are written on the new dictionary
tape as well as new additions and error-free changed records. If there
is an error in updating, the prior dictionary record is output on the
new dictionary tape. When the dictionary disk is being built, each
record that is output to the new dictionary tape is also output to the
indexed sequential disk file.
-34-
-------
Output
1.
Mount a scratch tape on SYS009 which will become the updated diction.
ary file tape.
2.
Mount a scratch tape on SYS008 which will contain the changes and
errors from the update program. This tape is printed using the
dictionary update message program (DICTMSG).
3.
Mount the dictionary file disk on SYS0l0. If it is not necessary
to build a dictionary file disk, then in the DOS job control cards
include the following: //UPSI lXXXXXXX. See DOS JCL for a de-
scription of this card (User Program Switch).
-35-
-------
DICTIONARY MESSAGE PRINTING PROGRAM (DICTMSG)
Purpose
DICTMSG lists the tape of changes and errors produced by the dictionary
maintenance program. The dictionary maintenance program processes three
types of dictionary modification requests: additions (new records),
changes (to subordinate data within existing records), and deletions (of
existing records). In addition, errors are detected and noted as to the
cause (i.e., invalid main term referenced, etc.). DICTMSG lists the mes-
sages created by the dictionary maintenance program and the dictionary
record(s) to which they apply.
Input
1.
Tape
DICTMSG requires that the dictionary maintenance change tape be
mounted on SYS006.
2.
Control Card
In some cases the listing of only one or two types of messages and
their corresponding dictionary records may be desired. When this
occurs, request data is inserted on an UPSI card at the beginning
of the program. This card sets certain bits in a communication
region of the supervisor which may be sensed by the program. Re-
quest data has meaning as follows:
//UPSI XXXXXXXX
12345678
A 1 in position four indicates an ad-
dition record will be printed.
A 1 in position five indicates a change
record will be printed.
A 1 in position 7 indicates a delete
record will be printed.
A 1 in position eight indicates an er-
ror record will be printed.
A zero (0) in any of the above positions
indicates that action is not requested.
-36-
-------
Processing
Each logical record is identified as either a message or a dictionary
record. Messages are classified as "requested" or "non-requested" de-
pending upon the request data described above. If the message is "re-
quested" (i. e., called for by an appropriate punch in the UPSI card),
it is printed and a switch is set indicating that the next dictionary
record encountered will be listed. When a dictionary record is identi
fied, the switch mentioned above is tested. If it is set, the dictionary
record is printed by the subroutine SBRFORM. Processing continues until
all logical records have been read. An end-of-file condition terminates
DICTMSG.
SBRFORM formats and prints the data in dictionary records read into the
calling program. The print format is described in the OUTPUT portion
below. Each term or block of subordinate data is identified and titled
before printing. The terms are indented and printed under the main term,
The main term is double printed (for emphasis) as are the titles for 'USE'
terms and 'SEE ALSO' terms. File/context and dictionary numbers are con-
verted to decimal prior to printing. When the end-of-data is encountered,
control is returned to the calling program.
Output
Each message will be printed on a separate line, A partial line of
asterisks will separate dictionary records. The dictionary records
will be printed in the following format:
(main term) (double printed)
AUTHORITY--(auth.)
FUNCTION--(fns.)
FILE/CONTEXT NUMBERS--
(f-c, f-c, f-c, f-c, f-c,...)
USE-- (double printed) (term
DICTIONARIES--(dictionary nos,)
or
RELATED TERMS--n (n=no. of related terms)
(rtl, rt2, rt3, rt4,...
rtn-l, rtn)
SEE ALSO-- (double printed)
(sal, sa2, sa3,...)
SCOPE NOTES--
LINE NO. n (text)
to be used in lieu of main term)
-37-
-------
A double space will separate the main term with its pertinent data from
the subordinate terms included in the record. Also a double space will
separate the dictionary record from the messages that reference it. An
example of the actual printout of a message and its dictionary record
follows:
~~~:~:::~:~::::::::::c:c:c:c:c:c:c:c:c:c
ECOTRIN
MAIN TERM HAS BEEN ADDED
ONE OR MORE ADDITIONS AND/OR DELETIONS
ECOTRIN
FUNCTION-- PD DICTIONARIES--l
FILE / CATEGORY NUMBERS--
1-5
USE-- ACETYLSALICYLIC ACID
SCOPE NOTES--
LINE NO.1 SS TRADE NAME HUMAN
LINE NO.3 SS MFG SMITH KLINE AND FRENCH
LINE NO.4 AMERICAN DRUG INDEX PHYSICIANS
DISK REFERENCE
-38--
-------
DICTIONARY PRINT PROGRAM (DPRINT3)
Purpose
This program enables the user to print selected portions of the dictionary
tape. The user may print an entire main term record or any combination of
the following parts of the record: authority of nomenclature*, dictionaries,
function, file/context codes, scope notes, see also references, related
terms, and use terms. He may either print every term on the dictionary
tape, or he may print selected terms. These options are controlled by a
user-prepared set of control cards.
* This field type is not generally carried now.
remains in the programs.
Coding to handle it
Input
This program uses input from both magnetic tape and cards.
1.
Tape
The updated dictionary tape is file protected and mounted on the
tape unit assigned as SYS~~7.
2.
Cards
The control (request) cards which control the printing are read
into the 2540 card reader assigned as SYSRDR. There are two basic
types of request cards.
a.
Dump Card
Th~ first type, the dump card, allows the user to print every
maln ~er~ record on the dictionary tape. One form of the dump
card lndlcates that each record is to be printed in its entirety.
The format of this request card is as follows:
DUMP / RECORD
-39-
-------
The other dump card form allows the user to print selected
portions of each record on the dictionary tape. The format
of this card follows:
DUMP / AN, DICT, FN , FC, SCOPE, SA, RT, USE
Where 'AN' indicates authority of nomenclature, 'DICT' indicates
dictionaries, 'FN' indicates function, 'FC' indicates file/con-
text codes, 'SCOPE' indicates scope notes, 'SA' indicates see
also references, 'RT' indicates related terms, and 'USE' indi-
cates use terms. Anyone or more of these options may be se-
lected. The options may be listed in any order.
b.
Term Cards
The other type of
used instead of a
ords. There must
be printed.
request is the term card. Term cards may be
dump card to print specific main term rec-
be a term card for each main term record to
The first form of the term request card will cause the entire
main term record that is specified to be printed. The format
of this card follows:
TERM / ABBOCILLIN j PRINT / RECORD
The second form of the term card will enable the user to print
selected portions of the main term that is specified. The for-
mat of this term card follows:
TERM / ANTI - INFECTIVES:
PRINT / AN, DICT, FN, FC, SCOPE, SA, RT~ USE
The options are the
as those on the dump request card.
There may be up to one hundred term request cards entered.
There may be multiple entries for anyone main term.
-40-
-------
Processing
Each term request causes the dictionary tape to be searched for the designated
main term. For a dump request, each dictionary record is considered a match.
If no match is found, an error message is printed and the program continues
with the next term request. When a match if found, control is transferred
to subroutine SBRFORM which prints the record and then returns control.
When all the requests have been processed, the files are closed and the job
is terminated.
Output
Output is handled by the subroutine SBRFORM.
printed in the following format:
Dictionary records will be
(main term)
(double printed)
AUTHORITY--(auth.)
FUNCTION-- (fn)
DICTIONARIES--(dictionary nos.)
FILE/CONTEXT NUMBERS--(f-c, f-c, f-c, f-c, f-c,...)
USE--(term) (double printed) (term to be used in lieu of main term)
or
RELATED TERMS--n (where n is the no. of related terms in the record)
(rtl, rt2, rt3, rt4, rtS,... rtn-l, rtn)
SEE ALSO--(couble printed)
(sal, sa2, sa3,...)
SCOPE NOTES--
LINE NO. n(text)
-41-
-------
Examples of the Printout of Spme Representative Dictionary Records follow:
ACETYLSALICYLIC ACID
FUNCTION-- PD DICTIONARIES--l
FILE / CATEGORY NUMBERS--
1-5
RELATED TERMS-- 5
ANALGESIC-ANTIPYRETIC, CNS DEPRESSANTS, NERVOUS SYSTEM DRUGS, NON-NARCOTIC ANALGESIC, SALICYLATES
SEE ALSO--
ECOTRIN, EMPIRIN, MEASURIN, SALACETIN
SCOPE NOTES--
LINE NO.1 SS GENERIC NAME HUMAN
LINE NO.4 MERCK INDEX NATIONAL FORMULARY
HELICON
FUNCTION-- PD DICTIONARIES--1
FILE / CATEGORY NUMBERS--
1-5
\
~
I
USE-- ACETYLSALICYLIC ACID
SCOPE NOTES--
LINE NO.1 SS GN-TN
LINE NO.2 SS GN-TN
LINE NO.4 SS MFG
LINE NO.4 REF MERCK INDEX
BARDASE VETERINARY
FUNCTION-- PD DICTIONARIES--1
FILE / CATEGORY NUMBERS--
2-5
RELATED TERMS-- 14
ASPERGILLUS ORYZAE ENZYME, ATROPINE SULFATE, AUTONOMICS, BARBITURATES, CARHOHYDRASE, CNS DEPRESSANTS,
ENZYME-ENZYME INHIBITORS, HYOSCYAMINE SULFATE, NERVOUS SYSTEM DRUGS, PARASYMPATHOLYTIC, PHENOBARBITAL
SCOPOLAMINE HYDROBROMIDE, SEDATIVE-HYPNOTIC, TROPANE ALKALOIDS
SCOPE NOTES--
LINE NO.1 SS GN-TN
LINE NO.2 SS GN-TN TRADE NAME VETERINARY
-42-
-------
SYNONYM TAPE PROGRAM (SYNTAPE)
Purpose
The master
terspersed
synonymous
dictionary tape produced by the DICT2 program contains, in-
with normal entries, a number of "use" entries, which contain
terms (e.g., "ASPIRIN use ACETYLSALICYLIC ACID").
The SYNTAPE (synonym tape) program examines the master dictionary tape
to extract these "use" entries and invert their form, so that the term
referred to by the synonymous word appears first in the record. After
extractionand inversion, the "use" entries are sorted in sequence by
the term referred to, thus bringing together all the variants which for
example, will be arranged alphabetically under that term. This sorted
tape, used in conjunction with the master dictionary file in DPRINT4
will permit the printing of a dictionary in which all such synonyms are
given under the term to which they refer as well as appearing in their
normal alphabetic sequence.
Input
The tape created by the dictionary maintenance program is mounted on
SYS007.
Processing
A record is read and the contents position is tested for the presence
of a "use" term in the record. If there is none, the next record is
read. If the contents position indicates there is a "use" term, it is
located and written on tape, followed by its length, its main term, and
the length of the main term. Then the next record is read. Processing
continues until the end-of-file.
Output
Output from SYNTAPE is written on a tape mounted on SYS008.
when sorted, will be used as synonym input to DPRINT4.
This tape,
-43-
-------
DICTIONARY PRINT PROGRAM (DPRINT4)
Purpose
This program enables the user to print the master dictionary tape in either
a publishable or a working copy format. The publishable copy is printed as
double-length pages with two columns per page. The working copy is printed
on single pages with two columns per page.
The user also may specify printing of the entire main term record or any
combination of the following parts of the record: authority of nomencla-
ture, dictionaries, function, file/context codes, scope notes, see also
references, related terms, and use terms.
In addition, he may select the specific dictionaries to be printed (e.g.,
he may choose to print only those terms belonging to the Adverse Reactions
Dictionary, Legal Dictionary, etc.).
Input
This program requires input from both magnetic tape and cards.
1.
Tapes
a.
The updated dictionary tape is file protected and mounted on
the tape unit assigned as SYS~~7.
b.
The sorted synonym tape is file protected and mounted on the
tape unit assigned as SYS~~6.
2.
Control Cards
Program control cards are read into the 2540 card reader assigned
as SYSRDR. There are three control cards required by this program.
a.
Copy Control Card
The first card, the copy control card, allows the user to print
the dictionary in either of the two possible formats. The for-
mats of this control card as as follows:
-44-
-------
COpy / PUBLI SH
COpy / ~.JORK I NG
b.
Dump Control Card
The second card, the dump control card, allows the user to dump
either the complete record for each main term or selected por-
tions of each main term. Following are the formats of this card:
DUMP / RECORD
DUMP / AN, DICT, FN , FC, SCOPE, SA, RT, USE
When dumping selected portions of the dictionary, any combina-
tion of the options below may be selected and in any order.
OPTION
DESCRIPTION
AN
DICT
FN
FC
SCOPE
SA
RT
USE
Authority of Nomenclature
Dictionaries
Function
File/Context Numbers
Scope Notes
See Also References
Related Terms
Use Terms
c.
List Control Card
The list control card allows the user to select the dictionaries
to be dumped. He may either specify that all dictionaries should
be included in the printed report or he may list the dictionary
mnemonics of those dictionaries to be printed. The formats of
this card are as follows:
LIST / ALL
LIST / XXX, XXX, XXX, XXX, ..., XXX
Where the XXX indicates a three character dictionary mnemonic, as
many dictionaries are allowed as will fit on one card.
The spacing on these cards is unrestricted; blanks between words
are ignored.
-45-
-------
Note the distinction between the
card and the dump control card.
tionary terms are to be printed.
of data to be printed under each
functions of the list control
The former controls which dic-
The latter controls the amount
term.
d.
Header Control Card
The header control card contains up to 60 characters of data
which will be used as a header for each page in the dictionary.
The format is simply the word HEADER followed by a slash (/)
followed by the header data.
e.
Restart Control Card
The restart card allows an operator to begin printing in the
middle of the dictionary. It is formatted as follows:
RESTART / PAGE=NNN, TERM=XXXXX
RESTART / TERM=XXXXX
Note that the page number is optional. If not included, pagina-
tion will begin at one. The term parameter must be included.
Both the dictionary and synonym tapes will be positioned to
begin at the specified term.
Processing
After initializing all switches, counters and working areas, and opening
the files, this program begins reading cards, searching for one each of
the three necessary kinds of control cards (copy control card, dump con-
trol card, and list control card). It will continue reading cards until
a system end-of-file card is encountered. The program will accept the
first three valid types it encounters, in any order. This means that
there may be any number of cards in the reader preceding the system end-
of-file. The program will simply accept the first three valid control
cards and ignore the remainder. If the program does not find three
valid cards, the job is terminated.
-46-
-------
Once the printing parameters have been determined, the program reads
records from the master dictionary tape and the sorted synonym tape
(which was produced by the SYNTAPE program). Dictionary terms which
meet the criteria specified in the list control card are printed along
with whatever elements of data were specified by the dump control card
and with the corresponding entries from the synonym tape, if any. If
a given dictionary term is not to be printed, it is ignored (along with
its associated synonyms, if any) and the next dictionary term is in-
spected.
This process continues until end-of-file is reached on the dictionary
tape. If, during the printing process, any records are encountered on
the synonym tape which do not correspond to entries on the dictionary
tape, there is an editorial error in the dictionary. Existence of a
record on the synonym tape in the form 'XXX use(d for) YYY', for ex-
ample, means that there must be a record on the dictionary tape in the
form 'YYY use XXX'. The editorial rules require that for such "use"
entries, 'XXX' be a valid dictionary term. The failure of the record
'XXX' use(d for) YYY' on the synonym tape to match with any entry on the
dictionary tape means that XXX does not exist as a main term in the
dictionary. Therefore, the cross reference, 'YYY use XXX' is invalid.
At the conclusion of the printing of the dictionary, these unmatched
synonym records are printed for inspection so that the dictionary edi-
torial staff may make corrections.
It is possible to execute this print program without having a synonym
tape available. This is done my mounting a scratch tape which has three
tape marks written at its beginning in lieu of the synonym tape on SYS~~6.
This will result in printing of the dictionary master tape but, of course,
without any 'used for' entries under any term.
Output
DPRINT4 will produce one of two types of printer output. These will vary
only in the length of the page (i.e., if a publishable copy is requested,
the page length will be equal to two printer pages. For a working copy;
only one page will be printed). Both types of output will follow the
format below:
-47-
-------
(main term)
AUTHORITY--(auth.)
FUNCTION--(fns.) DICTIONARIES--(dl, d2,...)
FILE / CONTEXT NUMBERS--(f-c, f-c, f-c, f-c,...)
USE--(term to be used in lieu of main term)
or
RELATED TERMS--
(rt! rt2
rt3 rt4
.. .)
SEE ALSO--
(sal
sa3
USED FOR--
(synonyml
synonym3
SCOPE NOTES--
XXXXXXXXXXX
XXXXXXXXXXX
sa2
sa4)
synonym2
. .. . ... .)
Each page will be divided into two columns, each following the above format.
If a main term record is split between columns (i.e., if the entire record
will not fit one column), a note will precede the continuation in the second
column. For example, ASPIRIN (cont'd.)
-48-
-------
~ A DICTI:JNARY
49
CALCI~~
CACCTuMODTNAT E
FUNCT1DN-- PDtSO DICTIONARIES- 1
F IL E I CONTE-X T CO DE S--
I I
--
CALCRET INI CONT .1
RElATED TERM -
MINERALSII-21
CALCTuM SACCHARATE
FUNCTION - PD,SD DICTIDNARIES -I
FILE I CONTEXT CDDES--
I I
RELATED TERMS--
BARBITURATESII-31
ETHYL AMINOBENZOATEI1-31
PAPAINI1-31
SEDATIVES-HYPNOTICSfl-2'
ENZYMESII-21
PANCREATIN(I-3'
PHENOBARBITALII-31
CALDECORT
FUNCTION - PO, SO
FILE I CONTEXT CUDES -
I-I
DICTIONARIES- I
R elATED TERM --
- PHARMACE"UTICAL AOJUN( 1-2)
RELATED TERMS--
ADRfNAl CORTICOSTEROW( 1 21
ANTIFUNGAL DRUGSII-21
CALCIUM UNOECYLr::NATE( 1 3.
HYDROCORTISONE ACETAII-31
N E OMY C I N~[nl'TrT=-31
ANTIBIUHCSII-Z'
CALCIUM SACCHARIN
FUNCT ION-- PO,5D DieT IONARIES--l
FILE I CONTEXT CODES--
I-I
S IEROIOS( 1-4'
--
RELATED TERM --
SALT-SUGAR SUBSTITUTII-21
CALDES ENE
FUNCTION-- PO, SO DICTIDNARIES--l
FILE I CONTEXT CODES--
I-I
CALCIUM SALICYLATE
FUNCTION-- PD,SD DICT[ONARIES
FILE I CONTEXT CODoS--
I-I
RELATED TERMS--
ANTI-INf-I:CT IVES( 1-2'
ANTIFUNGAL DRUGSI1-21
RElATED TERMS--
ANTIDIARRHEAl 1-2)
SALIC YLA TES C 1-4 t
C AL ENDUL A
FUNC1IUN-- PO,SO OIClIUNARlI:S--l
FILE I CONTEXT CODES--
I-I
CALCIUM SUCCINATE
FUNCTION-- PD,SO OICTIONARIFS--l
FILE I CONTEXT CDDES--
I-I
RELATE:D TERM --
ANTISEPTICSII-21
RelATED TERM --
ANALGES ICS-ANT IPYR ET ICI 1-21
CALFEV ITA
--ru-m::TTOlIJ-- PO, 50 0 IC T lUNI\R IE S--1
FILE I CONTEXT CODES--
I-I
CALCIUM SULFATE
FUNCTION-- PD,SO DICTIONARIES--l
FILE I CONTEXT CODES--
I-I
RElATED TERMS--
INDRGANICSII-41
PHARMACEUTICAL ADJUNII-21
Rt:LA I tD TERMS--
CAlCIUM PHOSPHATE 0111-31
FERROUS SULFATECI-3J
IRON PREPARATIDNSII-21
TH~HYORUCHLQR I( 1 3.
VITAMIN A PALMITATEII-31
VIIMlIN B ClJNlPltXll-lJ VITAP'fIN
V ITAM IN-MINERAL PREPARATIONII-21
ERGOCALCIFEROLII-31
INQRGANICSC 1-4)
VITAMIN AO-21
0(1 l J
CALCIUM SULFIDE
FUNCTIDN-- PD,SO DICTIDNARIES--I
FILE I CONTEXT COOES--
I-I
RELATED TERMS--
ANTISEPTICSII-21
CALGAGE-S
FUNCTION-- PO,SO OICIIUNARII:S -1
FILE I CONTEXT COOES--
I-I
I NORGANI C S 11-41
CALCIUM THIOSULFATE
FUNCTION-- PD,SO DICT!ONARIES--l
FILE I CONTEXT COOES--
I-I
RELATED TER... -
01 AGNOSTIC AIOSII-21
RELATED TERM --
ANT 1- INFECT IVES 11- 21
CALGLU CON
FUNCTION-- PO,SD OICTIONARIES--1
FILE I CONTEXT COOES--
I-I
CALCIUM TR ISJOIUM PENTETATE
FUNCTION-- PD,SD DICTIONARIES--l
FILE I CONnXT CODES--
I-I
REL A TED TERM s--
METABOLIC AGENTSII-21
REPLACEMENT SOLUTIONII-21
MINERALSII-21
R8..ATED TERM --
HEAVY METAL ANTAGONIII-2J
CAlHYDRATE
FUNCTION-- PO,SD OICTIONARtES--l
FILE I CONTEXT CODES--
I-I
CALCIUM UNDECYLENATE
FUNCTION-- PD,SD DICTIONARIES--I
F IL E I CONTEXT CDOES--
I-I
RelATED TERMS--
ANTI-IN FECT I VE S I 1-21
ANT IFUNGAL DRUGSII-2)
RELATED TERMS--
CALCIUM GLYCEROPHOSHII-31
DIURET ICSII-21
MERSALYL( 1-31
RENAL AGENTSII-21
XANTHINESII-41
CALCIUM LACTATEII-31
MERCURIALSII-41
MINERALSII 21
THEOPHYLL 1 NE 11-,31
CALCIWAFERS
FUNCTION-- PD,SD DICTIONARIES--l
F IL E I CONTEXT COOES--
I-I
CALIGES IC
FUNCTlON-- PO, SO DICTIONARIES--1
FILE I CONTEXT CODES--
I-I
RelATED TERMS--
CALCIUM GLUCONATEII-31
CALCIUM PHOSPHATE 0111-31
FAT SOLUBLE VITAMINSII 21
VITAMIN 011-21
V IT AMIN-M INERAL PREPARAT IONII-21
ERGQCALCIFEROLII-31
INORGANICSII 41
RELATED TERMS--
ANTI INFLAMMATORY AGENTII-2)
CALAMINEII-31
ETHYL AMINDBENZOATEII-3)
LOCAL ANESTHETICSII-21
M CPESOL HEXYLATEOII-31
CALCOTHEOBROM INE
FUNCTION-- PD,SD
FILE I CONTEXT CODES
I-I
OICTIONARIES--1
CAlINESE
FUNCTION-- PO,SO OICTIONARIES--1
FILE I CONTEXT CODES--
I-I
RelATED TERMS--
CARDIOVASCULAR ORUGSll 21
SALICYLATES 11-41
DIURETICSII 21
XANTHINESII-41
RELATED TERMS--
ANT AC I O-ADSOR B-OEMULlI- 21
CALCIUM CARBONATEII-31
GASTRO INTESTINAL AGENTII-21
INORGANICSII-41
MAGNESIUM CARBONATEII-31
MAGNESIUM TRISllICATlI-31
CAl CREOSE
FUNCTION - PD,SD DICTIONARIES--1
FILE I CONTEXT CDDES--
I I
CALIOB EN
FUNCTION - PD,SD DICTIONARIES--l
F IL E I CONTEXT CODES--
11
R elATED TERMS
_~EPTICSII-21
E XPEC TORANTSII-2)
CALCRETIN
FUNCTION POt SO DICTIONARIES -1
FILE I CONTEXT CODES--
I I
----
FDA D[CT IONARY
49
CALCIUM QU1N.A~_-
-------
SUBORDINATE TERM EXTRACTION (RELTERM)
Purpose
When terms are entered into the dictionary or when alterations are made
to the dictionary, some types of error checking are not feasible. For
example, it would be extremely time consuming to scan the entire dic-
tionary each time a main term is altered or deleted. Such a scan would
ensure that the modification will not leave some subordinate term without
an entry as a main term. RELTERM accomplishes this purpose. It ex-
tracts all subordinate terms of a particular type (e.g., related term,
use term, or see also term) and passes them to RELTRMI which validates
each term.
Input
1.
The dictionary tape is mounted on SYS007.
2.
A UPSI control card determines which types of terms will be extracted.
Its format is:
//UPSI 00000xxx
SRU
A I in the S position causes see also terms to be extracted.
A I in the R position causes related terms to be extracted.
A I in the U position causes use terms to be extracted.
If no control card is included, related terms will be extracted.
Processing
RELTERM scans each record to determine whether or not the requested term
types are present. If they are, each term selected is written out on tape
along with the main term to which it refers. Then the next dictionary rec-
ord is read. An end-of- file on the dictionary tape causes processing to
terminate.
-50-
-------
Output
The terms extracted and their associated main terms are written on a tape
mounted on SYS005. This tape, after sorting will provide the input to
RELTRM1.
-51-
-------
SUBORDINATE TERM VALIDATION PROGRAM (RELTRMI)
Purpose
After the subordinate terms have been extracted and sorted, RELTRMI per-
forms a validation against the dictionary. Any term not in the dictionary
as a main term will be detected and listed for correction.
Input
L
The dictionary tape is mounted on SYS007.
2.
The sorted extracted subordinate terms will be on a tape mounted on
SYS005.
Processing
RELTRMI processes each set of subordinate terms separately. That is, all
of the use terms are validated first (if there are any), then related
terms (if any have been extracted), and finally see also terms (if any)
are validated. The dictionary tape is rewound for each set of terms
checked. The sorted extracted terms are compared to the dictionary main
terms. If an extracted term is not located as a main term in the dic-
tionary, that subordinate term and the main term with which it was found
are written on the work tape. After the entire tape of extracted terms
has been processed, the work tape is rewound and printed.
Output
L
A work tape used to contain missing terms is mounted on SYS004.
This tape is later printed by RELTRMI.
2.
The printed listing of all subordinate terms which were not found as
main term entries in the dictionary. An example of the listing is
shown below.
-52-
-------
AOUADIAL WITH TESTOSTERONE
, OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
AQUADIOL WI TESTOSTERONE
BECOTIN WITH C
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
BECOTIN WI C
BENANSERNIN HYDROCHLORIDE
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
BAS
SEROTONIN BENZYL ANALOG
CAFFEINE AND SODIUM BENZOATE
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
CAFFEINE SODIO-BENZOATE
CAFFEINE SODIUM BENZOATE
CARDAMON SEED
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
GRAINS OF PARADISE
CHIMAPHILA
OCCURS WITH THE
FOLLOWING MAIN TERMS AS A USE TERM.
BITTER WINTERGREEN
GROUND HOLLY
PINE TULIP
PIPSISSEWA
PRINCES PINE
PYROLA
RHEUMATISM WEED
COMBEX WITH VITAMIN C
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
COMBEX WI VITAMIN C
DEXTROMETHROPHAN HYDROBROMIDE
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
D-METHORPHAN HYDROBROMIDE
EN-CEBRIN WITH FLUORIDE
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
EN-CEBRIN WI FLUORIDE
ERGOAPIOL WITH SAVIN
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
ERGOAPIOL WI SAVIN
GOODYS HEADACHE POWDERS
OCCURS WITH THE FOLLOWING MAIN TERMS AS A USE TERM.
GOODYS POWDERS
-53-
-------
THE MAIN FILE SYSTEM
-~,y-
-------
FILE MAINTENANCE SYSTEM
The heart of the FDA information retrieval system is the series of pro-
grams known collectively as the file maintenance system. It accepts
and processes the data from the documents to build a variety of infor-
mation files which are capable of being searched by the retrieval pro-
grams.
The Files
There are four closely inter-related, but physically distinct, files
of information involved in the file maintenance system. The most sig-
nificant of these is the master file. This file contains all informa-
tion from which the others are built, and is the basis of the retrieval
system. The contents of this file are described in detail elsewhere,
but are summarized here for reference. The master file contains a series
of individual records, one for each document which has been entered into
the system. Each individual record is divided into two portions. The
searchable section contains descriptors, subdescriptors, numeric values,
and similar data. It is that portion of the record which is used for
the actual search process. The second portion of the record is known
as "free text". Free text can be, as the name implies, any kind of
information one wishes to put into the record. It can be an abstract,
a summary of the original document, doctor's comments, or whatever
is desired. The free text can be treated as a series of "paragraphs"
or "segments" if desired. Different kinds of information may be re-
corded in different segments. For example, one may put the hospital
identification data in segment 1, diagnostic information in segment 2,
laboratory test data in segment 3, and so forth, depending upon the
user's requirements.
The master file is maintained in the IBM 2321 data cell (or its equi-
valent), a large-capacity direct access storage device. This permits
immediate access to any specific record in the entire file, analogous
to the manner in which one may turn to a particular page of a book
without having to peruse the preceding pages.
The records are maintained in this storage device in sequence by an
"internal document number" which is used solely for convenience in
computer processing. Each new document, as it is added to the file,
is assigned the next sequential internal document number. Thus, all
documents added to the master file are automatically stored after pre-
vious records.
-55-
-------
In order to permit easy reference to the machine-generated internal
number, a separate file (in effect a dictionary) is maintained. This
file contains cross references from the user's external document num-
ber (which may be alphabetic, numeric, or both, up to eight characters
in length) to the machine-generated internal number. It is called the
cross reference file.
A third file, called the inverted file, serves as a detailed index to
the master file. While the cross reference file described above pro-
vides a single entry for each document (i.e., one external document
equated to one internal number), the inverted file provides multiple
references to a given document. Each term (descriptor) used to describe
a document is entered in this inverted file, alphabetically with other
terms. Every term is followed by a list of document numbers (internal
numbers in this case) which identify the documents to which the term
has been applied. Under the term PENTOBARBITAL SODIUM, for example,
will be the list of documents to which that term has been applied.
Certain very common terms, such as DATE OF BIRTH, and the like, are
not indexed in this manner in the inverted file, since to do so would
create an excessively large volume of data. The use of these "common
descriptors" is described elsewhere. The inverted file, like the
cross reference file is kept on a high-speed direct access storage
device such as the IBM 2311.
The fourth file in this system is the ID (identification) file. This
is a special purpose file which contains, in effect, a subset of the
master file. It is written on magnetic tape (rather than on a direct
access device) and is in the same format. It contains user-selected
portions of the searchable section of the record for every document,
but contains no free text. It is used for certain types of searches
which are not feasible with the master file.
These four files represent the complete data base for the entire infor-
mation retrieval system:
1.
Master File
The master file contains all information, both descriptive terms
and free text, pertaining to all documents stored in the system.
2.
Inverted File
The inverted file is a subject index to the master file.
-56-
-------
MASTER FILE MAINTENANCE
IN PUT
CARDS
(OR)
M FED IT
DICTIONARY
ON DIRECT
ACCESS
DEVICE
LIST OF INPUT
DATA AND
TERMS MISSING
----
DI RECT AC-
CESS XREF
FILE
,,/
MFMAINT
D I R ECT
ACCESS
MASTER
FILE
~I NT TAPE
I} ADDITIONS
2) DELETIONS
3) CHANGES
4} ERRORS
5}I.D.FILE
REVISIONS
CROSS
REFERENCE
FILE UPDATE
MPRINTI
LISTING OF
MAINTENANCE
PERFORMED
I. D. FILE
REVISIONS
-57-
-------
3.
Cross Reference File
The cross reference file relates user assigned
the document numbers assigned by the system.
document numbers to
4.
Identification File
The identification file is a condensed form of the master file
which typically contains the common descriptor portions of the
master file records plus terms specifically tagged to become
ID file entries.
The programs which create these files are inter-related in such a way
as to ensure that, short of a deliberate stoppage, the files will al-
ways remain "in phase". That is, the programs, if allowed to run to
completion, will always update all four files in such a way that any
change in the master file will be correctly reflected in the other
three. It is not possible to change any of the other three without
first changing the master file.
The Programs
A series, of nine~ograms comprises this file maintenance system. They
are: MFEDIT" MFMAINT" UPDXREF" IDUPDATE" I FUPDATE" MPRII\!Tl" EDITBC"
DCELLDMP" and LOADCELL. Like the dictionary system, two of these per-
form the basic functions, while the remainder are, in effect, "satellite"
programs.
The two fundamental programs are MFEDIT and MFMAINT, master file edit
and master file maintenance, respectively. MFEDIT accepts card (or
card images on tape) input of additions, changes, and deletions to the
master file. All the various elements of information, such as the de-
scriptors, subdescriptors, and the like, are validated by checking them
against the disk file version of the dictionary. In addition, a large
number of internal consistency checks are made, ensuring, for example,
that every subdescriptor is in fact associated with some descriptor.
MFEDIT is designed in such a way that, if any portion of the record for
a given document is found in error, the entire document is rejected.
This ensures that only logically complete documents are passed onto the
master file maintenance program.
-58-
-------
INVERTED FILE MAINTENANCE
DSORT
IVUPDT
IVPRINT
IFPRINT
I NVERTED FILE
STATISTICS
INVERTED FILE
LISTING
-59-
DIRECT
ACCESS
INVERTED
FILE
-------
That program, MFMAINT accepts the validated data from MFEDIT, performs
the operations indicated (e.g., adds the new document, revises an exist.
ing document by adding or deleting elements of information, or deletes
a document from the file). Additional editing is performed by this
program. It prints error messages if one attempts to delete a non-
existent document, and the like. MFMAINT makes the changes directly
to the master file maintained in the data cell, and produces tape
records required to alter the status of the other files to bring them
in correspondence with the master file. These tapes are the input to
the three auxiliary file maintenance programs UPDXREF, I DUPDATE, and
IFUPDATE (update cross reference file, identification file update,
and inverted file update).
All three of these latter programs operate in essentially the same
manner. The new information is read in conjunction with the prior
version of the appropriate file and the indicated changes are made,
thus producing a current version of that file. In th~ case of the
cross reference and inverted files, both a tape and a disk version of
the updated file are produced. For the identification file only a
tape is produced.
The continued updating of the master file in the data cell will ulti-
mately result in the need for re-organization of the information in
that device and the necessity to generate back up on magnetic tape.
Programs are therefore provided to re-organize the data cell. DCELLDMP
(data cell dump) is a program which will read all information from the
data cell, in ascending sequence by internal document number, and copy
it onto tape. LOADCELL reverses the process, reading from the tape
and loading the data cell sequentially. This process (DCELLDMP fol-
lowed by LOADCELL) will optimize the file organization of the data
cell.
Since, in the process of executing DCELLDMP, every record in the file
must be read sequentially, there is an opportunity to make certain
changes to those records. It may be desirable, for example, to change
every occurrence of the descriptor ATOMIC PILE to NUCLEAR REACTOR. It
may be desirable simply to delete every occurrence of a given term.
This facility is provided by the block change feature of DCELLDMP. It
is possible to prepare cards to perform a delete or replace operation,
if desired. The program EDITBC (edit block changes) validates these
desired changes against the dictionary disk file and stores them for
use by DCELLDMP. As the latter program is reading the records out of
the data cell onto tape, it consults the list of block changes and
a~ters the master fi~e records accordingly. Thus, the resultant tape
fIle, at the conclusIon of DCELLDMP, is a completely current version
of the master file. In addition, the program generates tapes which will
serve as input to the inverted file and identification file update pro-
grams so that those files will also reflect the current status of the
master file.
-60-
-------
1.0. FILE MAINTENANCE
IDUPDATE
MPRINT2
LISTI NG OF
I.D. FILE
-61-
MPRINT2
LI STI NG OF
MAl NTENANCE
PERFORMED
-------
The last program in the file maintenance system series is MPRINTI, a
program which can be used to print master file records from a variety
of sources. It is normally used to print the changes, additions, and
deletions at the conclusion of a MFMAINT run, but it is also embedded
within LOADCELL to permit the printing of the entire file, if desired,
when reloading the data cell from tape.
-62-
-------
CROSS REFERENCE FILE
MAINTENANCE
DSORT
UPDXREF
XREFPRINT
LlSTI NG OF
CROSS REF
FILE
-63-
DIRECT AC-
CESS CROSS
REFERENCE
FILE
-------
FILE MAINTENANCE INPUT LANGUAGE
The file maintenance language uses a free form format to ease the burden
of indexing and key punching the data. The types of data recognized by
the file maintenance system are:
1.
Document Number
2.
Document Title
3.
Descriptors and their Context Codes (precise, common, trial, and
unverified)
4.
Subdescriptors
5.
Numeric Values
a.
b.
c.
d.
Single Values
Lower - Upper
Time Values
Date Values
Values
6.
Links
7 -
Free Text
The above data is entered into four types of card records:
1. Header
2. Title
3. Add
4. Delete
The contents of the cards vary somewhat depending upon whether data is
being added, altered, or deleted.
-64-
-------
Header
Columns
Columns
Column
Columns
1 5
6 - 8
9
10 - 80
Card Sequence Number
'HDR'
Blank
Header Card Data
Data pertaining to each document must be prefaced by a header card.
data entered in the header card determines what operations are to be
performed. Possible actions to be taken are:
The
1. ADD
2. REV (ise)
3. DEL(ete)
There are two document numbers associated with every record in the
master file. They are:
1.
External Document Number - nine characters maximum
Internal Document Number - 2563 -1 maximum
2.
The external number is the one assigned by FDA for the purpose of log-
ging, filing, etc. The internal document number is assigned by the
file maintenance program for use in the system's internal directories.
The external number may be alpha-numeric while the internal number is
a sequentially assigned binary number.
Title
Columns
Columns
Column
Column
1 - 5
6 - 10
11
12
Card Sequence Number
, TITLE'
Blank
Document Title - maximum of 24 characters
There are no restrictions, other than the maximum length, on the title
data. Since its primary purpose is for printing, it should be composed
of common computer printer graphics.
-65-
-------
Add
Columns
Columns
Column
Column
1 -
6
5
8
9
10
Card Sequence Number
'ADD'
Blank
Data to be added in free form indexing language.
Descriptors, subdescriptors, numeric values, data
values, time values, and links are entered via the
add card. Free text is also entered using the add
card, but it represents a special case in formatting.
Delete
The delete card makes possible the deletion of an entire document, de-
scriptors, subdescriptors, values, links, or free text.
A more detailed description of the file maintenance language follows:
1
Header Card
Columns 1 - 5 Card Sequence Numb er
Columns 6 - 8 'HDR'
Column 9 Blank
ADD: EXTERNAL No
Columns 10 - 80 REV: INTERNAL Security Yes
DEL INTERNAL
XOR : YES
ADD: EXTERNAL
\~en a document is being entered into the information retrieval system,
the HDR card must contain the ADD: EXTERNAL entry. If, for example, a
document numbered FDA 1234 were to be entered into the master file, the
ADD entry would appear as ADD: FDA 1234. Once entered into the master
file it will be assigned an internal number, but its identity as FDA 1234
will be maintained.
REV: INTERNAL
~en making changes to a document already in the master file, the record
must be referred to by the number assigned to it by the system. This
number is referred to as the INTERNAL number. If the system assigned
the number 60751 to FDA 1234, and we wish to revise the document's con-
tents, we must enter REV: 60751. A cross reference between external
and internal numbers is maintained by the system.
-66-
-------
DEL: INTERNAL
To delete a document from the master file, the user merely makes one
entry: DEL: INTERNAL document number. This entry results in the
deletion of all data pertaining to a document (all searchable terms
and free text). Deletion of selected portions of a document are con-
sidered revisions.
NO
SECURITY: YES
The SECURITY entry is optional. In the event one wishes to classify a
document, the security entry must be used. When used, the search pro-
gram will omit classified documents unless they are specifically re-
quested. That is, a classified document search is run.
XOR : YES
Before a document can be entered into the system, it is looked up in
the cross reference file. The document is rejected if it is already
present in the cross reference file. To override this restriction
the cross reference override entry in conjunction with ADD: EXTERNAL
must be used.
2 - Tit Ie Card
Columns
Columns
Column
Columns
1 - 5
6 - 10
11
12 - 80
Card Sequence Number
'TITLE'
Blank
Title Information
Title
Up to twenty four characters of title information may be stored with a
document. This data is not searchable and is only used for printing.
3 - Add Card
Columns
Columns
Column
1 -
6 -
5
8
9
Card Sequence Number
'ADD'
Blank
Columns 10 - 80
D : DESCRIPTOR; C:CXT ; LK:(1,2,.. .16) ; SUBDESCRIPTOR
LK: (1,2, . . .16) ;
-67-
-------
~
various forms of values include:
1 /:-
./.,
/; -
SV : DEC. or ~-
LV : \yr;OAT~L--
UV : NUMBER --
DV : MrvI-DD-YY
TV : HH.MrvI.SS
D : DESCRIPTOR
A descriptor may consist of from one to sixty four alphabetic or numeric
characters. A few graphics may not be used in forming descriptors as
they serve as punctuation in the indexing language. The reserved char-
acters are: semicolon and colon.
C : CXT
The CXT entry is provided to allow the user ~o indicate the context in
which a descriptor is used. In one case a r~~~~ay be a symptom while
in another case it might be a reaction. The' distinction can be made by
using the CXT code. The code itself is an abbreviation consisting of
from one to three alphabetic characters. The codes are stored as a
single byte binary number. Hence; there is provision for 256 unique
context codes for each data file in the system. In addition to the CXT
entry in the indexing language, provision has been made to append ad-
ditional context codes during the dict~onary validation phase of the
file maintenance system.
)
) /,'
" /-/-
The information retrieval system makes provision t~ relate a
to one or more descriptors by means of the LK (links) entry.
sixteen possible link entries for a given descriptor.
LK: (1,2,...16)
descriptor
There are
S : SUBDESCRIPTOR
The subdescriptor provision allows the indexer to refine or qualify the
meaning of a descriptor. Mass, for example, may be in either grams
or kilograms. The subdescriptor provides the ability to draw such a
distinction. The information retrieval system also makes provision to
relate a subdescriptor to one or more subdescriptors by means of the LK
(links) entry. There are sixteen possible link entries for a given
subdescriptor.
-68-
-------
SV : VALUE, LV : VALUE, UV : VALUE
Of the five types of values recognized by the system, three of them repre-
sent measures of various types of quantities and are entered in the same
format. Numbers can be entered in either normal decimal notation such
as the number 151.09 or floating point notation. The floating point no-
tation is a computer oriented form of scientific or what is commonly
referred to as engineering notation. It consists of a characteristic
and a mantissa in the reverse order. The number 1234.56 can be expres-
sed as .123456 x 104. In floating point notation, it would be expressed
as .123456+E4. In general, the format of a floating point number is:
+ D.DDD E + nn where D represents-a decimal digit and n
represents-the power of ten which the decimal point must
be moved to express the value being represented. Toler-
ances may be indexed by using the lower-upper value pro-
vision of the system. If used, both the lower and upper
limits must be specified.
DV : MM-DD-YY
or
DV : MM-DD-YYYY
Dates are indexed via the DV or date value entry. The system treats
dates differently than ordinary numbers since the year must precede
the month and day to effect accurate comparisons. If the YY format
is used, the system assumes 19YY. For dates outside the twentieth
century, all four digits of the year must be entered.
TV : HH.MM.SS
Time of day may be entered via the TV provision.
MM minutes, and SS for seconds.
HH represents hour,
Add Text Card
Columns
Columns
Column
Columns
1 - 5
6 - 8
9
10 - 80
Card Sequence
'ADD'
Blank
T : SEGNO
Number
T : SEGNO
This entry precedes the text data and is on a card by itself. SEGNO
may vary between 1 and 255. The text follows in the ensuing cards.
Text starts in column ten and can run as far as column eighty. There
-69-
-------
are no restrictions on the character set entered into the free text data.
The format for the text card is:
Columns 1 - 5
Columns 6 - 9
Columns 10 - 80
Delete Card
Columns
Columns
Column
1 5
6 - 11
12
Column
Columns 14
13
80
Delete Descriptor J
Card Sequence Number
Blank .'
Text Data
Card Sequence Number'
/
'DELETE' /
D (descriptor) /
S (subdescriptor)
V (value) .
L (link or links)
T (text segment)
Blank
D : DESCRIPTOR; S: SUBDESCRIPTOR ;
T : SEGNO
various forms of values include:
SV : DEC. or
LV : FLOATING PT
UV :
DV : MM-DD-YY
TV : HH.MM.SS
To delete a descriptor a 'D' is punched in column 12 and the D : DESCRIPTOR
entry is used. When a descriptor is deleted, its subdescriptors and/or
numeric values, and links are also deleted.
Delete Subdescriptor
To delete a subdescriptor, '(l 1 ~_'is punched in column 12. In this case,
both the D : DESCRIPTOR and the S : SUBDESCRIPTOR entries must be pre-
sent. The descriptor entry in this case informs the system under which
descriptor the subdescriptor to be deleted is stored. When a subde-
scriptor is deleted, its values and links are also deleted.
Delete Value
Values may pertain to a subdescriptor or a descriptor, depending upon
whether or not a subdescriptor is present. If a value is stored under
a subdescriptor, it pertains to the subdescriptor. If it is not stored
under a subdescriptor, it pertains to the descriptor. In deleting a
value associated with subdescriptors, which is the usual case, both
the D : DESCRIPTOR; and S : SUBDESCRIPTOR ; entries must be present
as well as the value to be deleted.
-70-
-------
When deleting a value not associated with a subdescriptor, only the
D : DESCRIPTOR entry is needed prior to the value to be deleted. When
deleting a lower-upper pair of values, both the lower (LV) and the
upper (UV) must be present.
When a descriptor is used to 'locate' a subdescriptor within a record
for the purpose of deleting (or adding) that subdescriptor, the de-
scriptor itself is not modified. It is referred to as a 'locator'
descriptor in the MFMAINT program. 'Locator' subdescriptors serve a
similar function in deleting or adding values.
Delete Links
Links can be deleted from a descriptor and/or subdescriptor in the
same manner as values.
-71-
-------
~~STER FILE ORGANIZATION
Introduction
The master file record organization in this system is more complex and
variable than is usual in most systems. This document describes that
record organization in detail, so that the description of the programs
which process it will be more easily understood.
Directories
The system maintains three directories for the purpose of retrieving
documents. These directories are built as documents are stored in the
device. ~~en a document is to be stored, it is assigned a sequential
internal document number. If a document is the last one in a strip,
its internal number is entered in the strip directory. If it is the
last one in a cell its internal number is entered in the cell directory.
The track directory is written on track zero of every strip. To retrieve
a document, the cell directory is searched to determine which cell it
may be in. Then the strip directory is searched to determine the strip
within that cell. Track 0 of this strip is then read and the track
directory for this strip is then searched to determine the track. This
track is read and searched for the internal number.
When a document is revised, it is tagged as deleted, given a new inter-
nal number, and written at the end of the file. These deleted docu-
ments are eliminated when the file is reorganized.
Directory Format
document
number
document
number
,. document
+-> .... +-> ~
~ 0) number ~ 0)
0),.0 0),.0
~ ~ ~ ~
bl);j bl);:!
0) ~ 0) ~
U1 U1
-----
document
number
+-> ~
~ 0)
0),.0
~ ~
bl);j
0) ~
U1
+-> ~
~ 0)
0),.0
~ E
bl);::S
0) ~
U1
-72-
-------
Continuation of Records
If the master record is longer than the capacity of a single track, it
is segmented and written on more than one track. Therefore, all records
contain a single byte field which serves as a continuation flag. If the
record will fit in the space left on a track, the continuation flag is
set to zero and the record is written in the remaining space. If it
will not fit, the continuation flag is again set to zero and as many
bytes as will fit are written on the track with the total length of the
record in the record length field. The number of bytes written on the
previous track are subtracted from the total length and the total
length moved into the record length field. The continuation flag is
set to one and the same procedure repeated. Thus, if the continuation
flag is '1', the record is a continuation of one from a previous track.
If the record length is greater than the space left on a track, the
record is continued on the next track. Using these three elements of
data (the flag, the record length, and the space remaining on a track)
the program is able to read back a complete logical record as required.
The continuation flag and additional record length fields are eliminated.
The record resides in core as logical entity with no extraneous data.
Description of Master Record
The master record is organized as follows:
1.
Continuation Flag
2.
Record Length
This two byte field is the total length of the master record.
3.
Internal Number
This three byte field is an internally assigned sequential number.
4.
Segment
This one byte field is zero if this record is a searchable section.
Otherwise, it represents the free text segment number (1 254).
5.
Tag
This one byte field represents the status of this record, that is,
whether the record has been deleted.
-73-
-------
10.
11.
12.
6.
1 Run Number
This two byte field is used for restart purposes.
7.
External Number
This nine byte field contains the external number assigned to the
document by the user.
8.
Security
This one byte field contains security information.
9.
File
This one byte field contains file identification.
Ti tle
This twenty four byte field contains title information.
Number of Pointers
This one byte field contains the number of pointers present in
this document.
Pointers
These seven byte fields are ordered to correspond with the alpha-
betic sequence of the descriptors in the document. Each one of
these fields has the following subfields. There is one pointer
for each descriptor in the record.
a.
Displacement
This two byte field contains a number which, when added to
the starting address of the master record, will give the
address of the descriptor to which this pointer pertains.
b.
Length
This one byte field contains the length of the descriptor.
c.
Rank
This one byte field contains a number which indicates the
relative position of each descriptor (i.e., first through nth)
in the document.
-74-
-------
d.
Type
This field contains a code to indicate whether a descriptor
is an expand term, main term, common descriptor, precise de-
scriptor, temporary precise descriptor, or I.D. file de-
scriptor.
e.
Number of Categories
This one byte field contains the number of categories associated
with this descriptor.
f.
Number of Qualifiers
This one byte field contains the number of qualifiers associated
with this descriptor.
13.
Descriptors
The descriptors are stored as variable length entries, maximum length
being 64 bytes. Following the descriptor there can be from 1 to 255
one byte categories associated with the descriptor. Following the
categories there may be from 1 to 255 qualifiers. A qualifier is
an element of information used to 'modify' or increase the specificity
of a descriptor. Each qualifier is composed of two parts: a sub-
descriptor and up to 255 numeric values and/or links associated
with the subdescriptor. The qualifier is organized as follows:
a.
Length
This one byte field contains the length of the subdescriptor
plus two.
b.
Number of Values/Links
This one byte field contains the number of
there is a link field present, it contains
plus one.
values present. If
the number of values
c.
Subdescriptor
This variable length field contains the subdescriptor.
length of a subdescriptor is 64 bytes.
Maximum
d.
Code
This one byte field indicates the type of data that follows,
i.e., whether it is a link or value. If it is a value, it
also indicates whether it is a single value, lower value,
upper value, time value, or date value.
-75-
-------
e.
Value
This five byte field contains the value converted to an internal
format for compactness.
14.
Note that either the subdescriptor or
(i.e., null subdescriptor and/or null
In the case of null subdescriptor the
associated with the descriptor.
the values may be missing
values are permissible).
values that follow are
Format examples of each record in the files are given
entification file record is in the same format as the
record except that no text is allowed.
below. The id-
master file
-76-
-------
MASTER FILE RECORD
A.
Searchable Section
LENGTH
1 byte
2 bytes
3 bytes
1 byte
1 byte
2 bytes
9 bytes
1 byte
1 byte
24 bytes
1 byte
2 bytes
1 byte
1 byte
1 byte
1 byte
1 byte
1 to 64 bytes
1 byte
1 byte
1 byte
0 to 64 bytes
1 byte
2 bytes
1 byte
4 bytes
CONTENTS
Continuation Flag
Record Length
Internal Number
Segment Number
Tag
Run Number
External Number
Security Code
File Number
Title Information
Number of Pointers
Displacement
Length of Descriptor
Rank of Descriptor
Type of Descriptor
Number of Context Codes
Number of Qualifiers
Descriptor
Context Code - 255 Allowed
Subdescriptor Length + 2
Number of Values/Links
Subdescriptor
Link/Value Code
Link
Value Characteristic
Value Mantissa
-77-
Pointer to
Descriptor
Information.
255 Allowed
-------
B.
Free Text Segment
LENGTH CONTENTS
1 byte Continuation Flag
2 bytes Record Length
3 bytes Internal Number
1 byte Segment Number
1 byte Tag
1 byte Text Line Length + 2
1 byte Text Line Number
1 to 71 bytes Text Line
1 byte Text Line Length + 2
1 byte Text Line Number
1 to 71 bytes Text Line
-78-
-------
INVERTED FILE RECORD
LENGTH
CONTENTS
2 bytes
Entire Record Length
2 bytes
Binary Zeros
2 bytes
1 byte
String Number
Continuation Flag
3 bytes
Highest Document Number this String
2 bytes
Total Document Numbers this String
6 bytes
Date Last Addition
6 bytes
1 byte
1 byte
l
-------
CROSS REFERENCE RECORD
LENGTH CONTENTS
9 bytes External Document Number
3 bytes Internal Document Number
-80-
-------
MASTER FILE MAINTENANCE EDIT PROGRAM (MFEDIT)
Purpose
MFEDIT reads in the additions and/or changes to the master file and valid-
ates them for correct operations, sequences, and contents. Valid inputs
are then reformatted and written on tape for use in the master file update
program MFMAINT. Invalid inputs are printed out with explanatory error
notes for correction of these inputs.
Input
Input to MFEDIT may be from cards ~r tape. Tape input must be card images,
blocked by a factor of ten in the following format:
Columns 1 through 5 must contain a sequence number. Operation to be per-
formed must start in column 6 and must immediately be followed by at least
one blank. The first non-blank character following is the start of the
data required for the operation. The data may be continued from one card
to the next by leaving the operation field of the next card blank.
For specific types of acceptable data refer to the section on file main-
tenance input language.
Processing
A data set comprises all the modifications that pertain to
ment. Each data set (i.e., all records between HDR cards)
completely before the output records for that data set are
Any errors encountered during the processing are printed.
a specific docu-
is processed
written on tape.
A record is read and scanned for errors, both in format and content. Such
things as descriptor validity and data sequence (e.g., context codes may
only follow a descriptor) are checked. Each verified piece of data be-
comes part of a reformatted output record.
-81-
-------
When all records of a given data set have been processed (i.e., when the
next HDR record is encountered) a test is made to determine whether any
errors have been found which could cause an invalid modification to the
master file document. If there have been none, the output records are
written on tape and the next data set is processed. If major errors are
found, the entire data set is rejected. Processing terminates when the
end of the input file is reached.
Output
1.
Tape
A tape of changes to the master file will be written on SYS~~4.
This tape will provide the input to MFMAINT.
2.
Listing
A listing of all errors encountered during the edit will be printed
along with diagnostics indicating the probable cause of the error
and the data in which it occurred.
-82-
-------
MASTER FILE MAINTENANCE PROGRAM (MFMAINT)
Purpose
After the change data for the master file has been edited (MFEDIT), it
is submitted to MFMAINT for processing. MFMAINT updates the master
file contained on the IBM data cell and creates tapes of update infor-
mation for the inverted file, the cross reference file, and the 1.0.
file. These tapes are processed by later programs. In addition errors
in data requesting modifications to the file are detected and a message
indicating the nature of each is printed.
Input
The input to update the master file is arranged in MFEDIT to a format
which is acceptable as input to MFMAINT. A logical input record has
the following format:
BYTES
P I
2
3 - 4
5 - n
CONTENTS
Length of logical record
Operation to be performed with the following data
Length of data following
Variable length data to be operated upon
Example: If it is desired to add a descriptor 20 characters long to a
certain document, the input record to MFMAINT will be as follows:
BYTE~
o - 1
2
3 - 4
5
6 25
CONTENTS
X'0019'
X'01' (add descriptor)
X'0015'
Type of descriptor
Actual descriptor
Since most of the input records will be relatively short in length, they
are blocked out of MFEDIT. MFMAINT deblocks and processes them one logi-
cal record at a time.
-83-
-------
Processing
This program performs three basic types of operations on the master file.
These are:
1.
ADD (addition of a new document to the file)
2.
DELETION (complete removal of a document from the file)
3.
REVISION (alteration of some elements of information pertaining to
some document already in the file)
The detailed specifications for preparing these inputs is given in the
documentation for MDEDIT (master file edit program). The operation pro-
ceeds along the following lines. Every alteration of whatever nature
is identified by a HDR (header) record, in the following pattern:
1. HDR ADD external document number (addition)
2, HDR DEL internal document number (deletion)
3. HDR REV internal document number (revision)
HDR ADD records are followed by 'trailer' records, all relating to the
document identified by its external number in the HDR ADD record. These
trailer records contain the descriptors, subdescriptors, free text, and
other elements of data pertaining to that document.
HDR DEL records may not be followed by any trailer records. A HDR DEL
record will cause the indicated document to be completely purged from
all the system files.
A HDR REV record is used to make changes to existing records. Like HDR
ADD records, HDR REV records are followed by trailer cards. These trailer
cards indicate the specific revisions to be made in the master file record.
This program gets each logical record in turn from the input tape (the
tape generated by MFEDIT), determines what kind of record it is (HDR
or trailer record) and performs the necessary operations. That is de-
pending upon the nature of the input, the program will add a new docu-
ment to the file, delete an existing one, or alter an existing one.
The master file record, contained on the large capacity direct access
file, is altered at this time. Records necessary to alter the status
of the other three files (I.D. file, cross reference file, and inverted
file) are also generated at this time. In addition, a tape record of
the changes made, and any errors encountered, is written for later printing
-84-
-------
Output
A.
Tape
1.
Update for the inverted file will be written on SYS~~5.
sorting, this will provide input to IVUPDT.
After
2.
Update to the cross reference file will be written on SYS~~8.
After sorting, this will provide input to UPDXREF.
3.
Update to the identification file will be written on SYS~~7.
This will provide input to IDUPDT. No sorting is required.
4.
Data which will be listed by MPRINTI is written on SYS~~6.
B.
Direct Access
The update data will be written directly into the data cell or
other large capacity direct access device.
-85-
-------
MASTER FILE PRINT PROGRAM (MPRINTl)
Purpose
MPRINTl lists the tape of changes and errors produced by the master file
maintenance program. There will be five types of records: additions (to
the current master file), changes, deletions, errors (e.g., duplicate
external numbers encountered), and ID. file revisions. The last type
of entry will not be listed by MPRINTl. Instead it will be transferred
to a tape for later processing.
Input
Input to MPRINTl will be a tape of master file records mounted on SYS~~6.
Processing
MPRINTl will process the master file entries in order by type (e.g., ad-
dition, change), listing the entire master file record acted upon. On
the first pass, the addition records will be listed and the other four
types written on work tapes. On the next three passes, change records,
deletion records, and error records will be listed. The rD. file re-
vision tape will be saved for later processing. Titles indicating the
type of modification being listed will be printed at the top of each
printer page. The output format of the records is discussed in the
OUTPUT section. Processing will terminate when an end-of-file is en-
countered on the error tape.
Output
Four tapes will be created by MPRINTl. Three of these are work tapes and
will be read back in by MPRINTl during later processing.
-86-
-------
1.
A work tape containing change records will be written on SYS~~5.
A work tape containing delete records will be written on SYS~~4.
A work tape containing error records will be writton on SYS~~7.
2.
3.
4.
A tape containing ID. file revisions will be written on SYS~~8.
This tape, after sorting, will provide input to IDUPDT.
-87-
-------
MASTER FILE MAINTENANCE LISTING 02/20/68
-~-~-_.- ADDIT IONS ----------------_..- ---- - --
PAGE I
----- -.---- - - -
INTERNAL NUMBER I,EXTERNAL DOCUMENT NUMBER 0030010 ,FILE I,IDfNTIFICATION 00300101
01 : GEI\I; t.): ()()()OnOOO; '): ~~--S:-~-oV: 06-0 1-6b; 5: NY; '~-fDP:;- DV:dT=-1-Cf-67 ~-- - S: N[j-; ----~'J::-IF r; -S :N$rff"f----S:~EC-; ---S-: 01 f -. -- -------~----
D~:URTICARIA; S:AR; n3:SKIN/DERM/HYSN; D4:HYSN/ATOP; D5:IMFERON; C:l; S:IRON PREPARATIONS; S:LAKESIDE; S:INJ; S:CC; SV:.2E4;
S:IM; 5:STPT: nV:Ol-10-67; S:STOP; DV:OI-10-67; S:DAYS; SV:.IE1; 06:2930; D7:NI0787; lK:(9); DS:HEMATOLOGIC AGENTS; C:2;
D9: IRON PfITlJARATIONS; C:2 - -------- ---- _n- - ----- -~--- - -- - -- ----------
INTERNAL NUMRFR ?,EXTERNAL DOCUMENT NUMBER 0030020 ,FILE I,IDFNTIFICATIDN 00300201
Dl : GEN; S: nnOOOOfJO; S: ITXI',r;-S:-IfD;-DV: 06- 29-2 5; S: NY; S: DR :--DvToF--03=-6-7; - S: NO; c; V-:-~ IE 1; $:NSl:P:;- S: R-fC; - S: HT; - n SV:~-65r:2;-
S:WT; SV:.3E3; S:02; D2:URTICARIA; S:AR; 03:SKIN/DERM/HYSN; D4:HYSN/ATOP; D5:RENOGRAFIN; C:l; S:ROENTGENOGRAPHY; S:SQUIBB;
5:INJ; 5:CC; SV:.35E5; S:IV; S:STRT; OV:OI-03-67; S:STOP; DV:Ol-03-67; S:DAYS; SV:.IE1; D6:X931; D7:NI0040; LK:(9);
DR: IODINAHn COMPOS; C:4; DQ:ROENTGENOGRAPHY; C:2 ------~____~n_~____---______n_--- ---------~~----
INTERNAL NIJMBER 3,EXTERNAL DOCUMENT Nur~BER 0030030 ,FILE 1, IDENTIFICATION 00300301
01: GE"1; 5: 0061155 R; S: SEXF ; S: BO; DV: 01-16-67: S: NY; S: o-ff;- -O\! ioF 16-67; --$: NO; - S Vi- .H'! ; --S-if.fSER-;_n_$:-lfEC; -S-:HTf--s'1:~6E~
S:WT: SV:.13F3; S:OI; 02:TREMOR: S:AR; D3:NER/CNS/CRN: 04:RENQGRAFIN; C:l; S:ROENTGENOGRAPHY; S:SQUIBB; S:INJ; S:CC;
SV:.2E6; 5:IV; S:STRT; nV:Ol-16-67; S:STOP; DV:Ol-1b-b7; S:OAYS; SV:.IE1; D5:X931; D6:NI0040; LK:(9); D7:IOOINATED COMPOS;
C:4; 08:RClf:NTGENOGRAPHY; c:2 -------------- - - - - - - - - - - - - ---- ----
INTERNAL NUMRER 4,EXTERNAL DOCU~f:NT NUMBER 0040020 ,FILE I,IOENTIFICATION 00400201
nl: GFN; S: 00 11 8 155; S: SEX'1;-i):f\O~-OI-14-22; S: ARK; S: DR; DV: 6i:"09-67;--- s-:f\Jo-;--sv:.jlT;--s-£f\lSE-R;-S-:REC;---s:--HT;-SV: .6-8('2;-
5:WT; SV:.15E3; S:OI: nZ:RASH MAC PAP; S:AR; D3:SKIN/DERM/f~Y; 04:HYSN/NEC; D5:PHENOBARBITAL: C:l; S:SEDATIVES-HYPNOTICS;
D6:I~POSS; S:TAB; S:MG; SV:.18F6; SV:.12F4; S:OKAL; S:STRT: DV:OI-0S-67; DV:OI-06-67; S:STOP; DV:Ol-0q-67; DV:OI-06-67;
S:DAY5; SV:.2El; SV:.IEl; 07:3240; OS:MORPHINE SULFATE; C:l; ;S:NARCfnrt-Af{tiLGESTCS-;--09:7S17;--DI0-:-BARBITTfRATi::S;C:4 ---
Dl1:NERVOUS SYSTEM DRUGS; C:2; 012:SEDATIVES-HYPNOTICS; C:2; DI3:NARCOTIC ANALGESICS; C:2; 014:0PIUM ALKALOIDS; C:4
-~-~-- -~ ----
-------~~~--- ~--- -- -~--~_._- ~ -- ----
-------~-
\
~
~
\
INTERNAL NUM3fR 5,EXTERNAL DOCUMENT NUMBER 00R0020 ,FILE I,IDf:NTIFICATION 00800201
DI:GE"1; S:63024046; S:SEXF; S:80; OV:OO-OO-ll; S:NY; S:OR; nV:01-19-b7; S:ND; SV:.1El; S:NSER; S:REC; S:OI; 02:0YSPNEA
S:~R; 03:RES/FUNC; D4:NICDTINIC ACID; C:l; S:VITAMIN B COMPLEX; S:TAB; S:MG; SV:.5E5; S:ORAL; S:STRT; DV:Ol-1Q-67; S:STOP;
DV :01-1 9-67; S:f1AYS; SV: .1E I; D5:368~D6: VAsdOTLATORS; C:2;07: V ITAMfi\! a COMPLEX; -(Ti- -- - ---------- - -_u_------
INTERNAL NUMBER 6,EXTERNAL DOCUMENT NUMBER 0080020 ,FILE I,IOENTIFICATION 00S00202
Dl: GEN; s: 61024046; S: SEXF; S: so; DV: 00-00-11; S: NY; S: OR; DV-:-(rl-19-b7; -si-N~--C;V:-:-lET;-s-: NSER;--S:REC;---S:or;~--~--~-~
02:RASH MAC; S:AR; n3:SKIN/DERM/CUT; 04:HYSN/NEC; D5:NICOTINIC ACID; C:l; S:VITAMIN B COMPLEX; S:TAB; S:MG; SV:.5E5; S:ORAl;
S:STRT; OV:OI-19-67; S:STOP; DV:01-19-67; S:DAYS; SV:.1El; 06:36S9; D7:VASOOILATORS; C:l; D~:VITAMIN B COMPLEX; C:2
----------------- -- ----------- - --...-- . -- - .---. - .-----------
INTERN.AL NUM'~FR 7,EXTERNAL DOCUMENT NUMBER 00S0030 ,FILE I,IDENTIFICATIClN 00S00301
Dl:GEN; S:00005914; S:SEXM; S:BO; OV:OS-15-15; S:NY; S:OR; DV:02-00-65; S:NO: SV:.2El; S:SER; S:A/S; S:WT; SV:.14E3;
S:02; D2:LYMPHADENO; S:AR; D3:HATI[YMIGLN; D4:HYSN/SST; D5:0ILANTIN;-t-:f;~s-:iHi'ri[bNIlOLSANTS; --:":IYARKE-DAVfs; . 06:IAREM; S:CAP;~-
S:MG: SV:.1E6; SV:.q~5; S:ORAL; S:5TRT; DV:OO-OO-61; S:5TOP; DV:00-00-67; S:OAYS; SV:.219E4; S:TAB; 07:3531; OS:N16480;
LK:(q); Oq:PHFNUBARBITAL; C:l; S:SEDATIVES-HYPNOTICS: 010:ANTICONVULSANTS; C:2; Dll:HYOANTOINS; C:4; D12:BARBITURATES; C:4;
D13:~FRVOUS SYS.1EM ORUGS;---ET2;Df4:S-EOATTvE-S:"'HYPNOTICS; C:2--------------_d ----. - --.--- -.--------.-- -_.~--- .-.--.- - - - -~ - -
INTERN~L ~U~BfR a,EXTERNAL DOCUMENT NUMPER 0090090 ,FILE I,IOENTIFICATION 00900901
01 : GEN; S: ()()() 19061: 5 -:si=X~f;~ S:Rr:'f;--- IYv:oo=oo=- rs;----STNY ;--S-: OR; ---[) v:IT':: 6-6- 6 b;- - S: ITo;- -S-V:. -2El;--s:-lifsER ;--------s:R-EC; ..s: HT; . S v: ~ 67E2;--
S:WT; SV:.149E3; 5:01; 02:HEMATURIA; S:AR; D3:GU/UT/URN; D4:HMRG; D5:COUMADIN; C:l; S:ANTICOAGULANTS; S:ENOO; S:TAB; S:MG;
--~ SV :---!51:4; S: 0-'3.~,=--~TR !J_------Q~.:.~g~QB-6A ; S: STOP; DV .:J..1-05-66~:Q.~~.?_;---~-=.....o_E2..: 0_6_: i..~~Q ;_~_~:_N.9..?_?.!_S;- _Ll~u: (9); 08: ANT ICOAGULANTS;
C:? - - .
INTERNAL NU~BER 9,fXTERNAL DOCUMENT NUMBER 0130030 ,FILE I,IOENTIFICATION 01300301
-----111: GEt{; S: ()()()[)oooo:s: SExF;-------s:-fjo;---1)lr: 05-06-21;S--:-TE X; 5: OR;--DV :oFTcf'::-i;7-; - - -s:-Nb;---sV: .lEl ;-S:NSER;- S:REC; -S:wT; -- SV: ~154o-
5:02; 0?:?ASH; S:AR; D3:SKIN/DERM/FRY; D4:HYSN/NEC; D5:DILANTIN; C:l; S:ANTICONVUlSANTS; S:PARKE DAVIS; S:CAP; S:MG;
SV:.3F6; S:DRAL; S:STRT; OV:01-19-67; S:STOP; DV:01-23-67; S:OAYS; SV:.5El; 06:4469; 07:0; LK:(9); D8:NI64S0; LK:(9);
DQ-;-,fNtTcoHI1Dr SMn 5; C:;>; 010: HVITAi\fTOfN~; C: 4 --- ---------- ____n___~~_---------------------- m --- --_u_- --
__~E R r-J~L~I_LIf'qFP 1.°-,- EX TE--,R~A!-__DSJ_~Y~NT Nl2.r'.!3J:~~o_goo 4~£l!:. E. .GJ OENTJF_~ <::",A II_QN- 91300401
-88-
--- ----.- ----
. ------------- ---------~------ - --.----------
-------
ID.
FILE UPDATE PROGRAM ( IDUPDT)
Purpose
As a result of a master file update run, certain 10. file descriptors may
be added, changed, or deleted from the master file. This program updates
the 10. file so it will correspond with the status of these descriptors
in the master file.
Input
Three tapes will be acceptable as input to IDUPDT.
They are:
1.
The tape containing the prior
10. file will be mounted on SYS005.
2.
The tape containing new additions to the
on SYS007.
10. file will be mounted
3.
A third tape which contains revisions to already existing
records will be mounted on SYS004.
10. file
Processing
IDUPDT reads records from the revisions and merges them with the records
from the prior 10. file. When the document numbers are equal and a
delete tag is set In the revision record, the prior 10. file record
thus identified is written onto the print tape. This can be printed
using MPRINT2 for a listing of deleted 10. file records. Otherwise,
the input record is written on the output tape. When an end-of-file
is read from the revisions tape, records are read from the additions
tape and placed at the end of the new 10. file tape until end-of-file
is reached on the additions tape which terminates the program.
-89-
-------
Output
1.
The new
10. file tape will be written on SYS00l.
2.
Any deleted records will be written on a tape mounted on SYS006
This tape may be listed by MPRINT2, a slightly modified version
of MPRINTl.
-90-
-------
UPDATE CROSS REFERENCE FILE (UPDXREF)
Purpose
UPDXREF updates the cross reference file for the master file. The cross
reference file is the ordered listing of the external/internal document
numbers resident on disk. The pairs are in order according to the col-
lating sequence of the external number. The master file maintenance
program uses the cross reference file as a validity check on modifica-
tion requests to the master file. For example, a check can be made to
determine whether or not a document which is to be added is, in fact,
already in the file.
Input
UPDXREF uses two tapes and a disk as input.
They are:
1.
The sorted tape created by MFMAINT consisting of update to the cross
reference file will be mounted on SYS~~4.
-'<} j,
The tape containing the prior cross reference file will mounted on
SYS~~7.
2.
3.
The cross reference disk is mounted on SYS~18.
Processing
The two tapes are read and the records are merged in order by external
document numbers. Since the tape created by the master file maintenance
program will contain both additions and deletions, these records are
tested for type and the indicated functions performed (i.e., if a rec-
ord specifies delete, the record referenced is located and deleted).
The tape containing the prior cross reference file is read and acted
upon. Processing continues until an end-of-file is encountered on
both tapes.
-91
-------
Output
UPDXREF creates two types of output.
They are:
1.
Tape
The tape containing the new cross reference file will be written on
SYS008. This tape may be listed by XREFPRNT.
2.
Disk
The direct access version of the cross reference file will be written
on SYS0l8.
-92-
-------
CROSS REFERENCE PRINT PROGRAM ( XREFPRNT)
Purpose
This program provides a listing, for maintenance purposes, of the cross
reference file.
Input
The cross reference tape file is mounted on SYS004.
Processing
XREFPRNT lists external/internal numbers in eight columns. The external
number is compared to the previous external number processed, since dup-
licate external numbers may occur if the indexer has overridden the dup-
lication option in MFMAINT. All internal numbers are listed below the
external number to which they are related. Processing terminates when
the end-of-file is encountered on the input tape.
Output
A printed listing of the cross reference file is output.
this listing is shown below.
A sample of
-93-
-------
CROSS-REFERENCE LISTING 11/01/67
fXTE:,NAL EXTER\Jl\L EXTERNAL EXTERNAL EXTERNAL EXTERNAL EXTERNAL EXTERNAL
1NTERNl\L 1NT~RNAL INTERNAL INTERNAL INTERNAL INTERNAL INTERNAL INTERNAL
1 29 399 4999 59999 699999
I 31 61 2641 10321 18001
2 32 62 2897 10577 18257
3 33 63 3153 10833 18513
2 39 499 5999 69999 799999
4 34 64 3409 11089 18769
5 35 65 3665 11345 19025
6 36 66 3921 11601 19281
3 49 599 6999 79999 899999
7 37 67 4177 11857 19537
8 38 68 4433 12113 19793
9 39 69 4689 12369 20049
4 '59 699 7999 89999 999999
10 40 70 4945 12625 20305
11 41 71 5201 12881 20561
12 42 72 5457 13137 20817
J 5 69 799 8999 99999
13 43 73 5713 13393
~ 14 44 74 5969 13649
t 15 45 75 6225 13905
6 79 899 9999 199999
16 46 76 6481 14161
17 47 77 6737 14417
18 48 78 6993 14673
7 89 999 19999 299999
19 49 79 7249 14929
20 50 80 7505 15185
21 51 81 7761 15441
8 99 1999 29999 399999
22 52 337 8017 15697
23 53 593 8273 15953
24 54 849 8529 16209
9 199 2999 39999 499999
25 55 1105 8785 16465
26 56 1361 9041 16721
27 57 1617 9297 16977
19 299 3999 49999 599999
28 58 1873 9553 17233
29 59 2129 9809 17 48 9
30 60 2385 10065 17745
-94-
-------
INVERTED FILE UPDATE PROGRAM (IVUPDT)
Purpose
Whenever the master file is updated by MFMAINT, a tape of the modifications
which influence the inverted file is created. This tape is used by the
inverted file update program to alter the inverted file tape and the in-
verted file disk. In this manner, the inverted file will always reflect
the current status of the master file.
Input
Input consists of two tapes:
prior inverted file.
the sorted inverted file update and the
1.
Inverted file update tape is mounted on SYS~~4.
2.
Prior inverted file tape file is mounted on SYS~~6.
Processing
The record changes (inverted file update from the master file update pro-
gram) will be organized alphabetically by main terms, and, if more than
one change is requested per main term record, by document number. The
two tapes (prior inverted file and sorted update) will be processed con-
currently and comparisons made between the main terms of each record on
either tape to determine whether or not the prior inverted file record
being scanned is to be altered. These comparisons will have the follow-
ing significance:
1.
If the update main term is low, an inverted file record will be
created for this term and update records immediately following
which refer to this new record will be processed. The entire
new record thus created will be output.
-95-
-------
2.
If the update main term is
prior inverted file tape.
be output unchanged.
high, it refers to a later record on the
Then the prior inverted file record will
3.
If the update main term is equal to the prior inverted file main
term, the type of alteration is checked. If it is a deletion of
the entire record, a message indicating the nature of the change
is written on SYSLST and the next record from both input tapes
is read. If it is not a 'delete all' request, the change requested
is made, all indicated modifications are made, and the next up-
date record is read. When all records referring to this record
have been processed, the altered record is output.
As soon as an end-of-file condition is encountered on either tape, the
main term area of that tape file is filled with high values and
the file is closed. In this manner, any compare will indicate that
the other tape record is low and is to be processed first. When an
end-of-file condition occurs on both tapes, the file is closed and pro-
cessing terminates.
Output
Output consists of one tape file, one disk file, and a listing of any
errors encountered during the update run.
1.
Tape
The updated inverted file tape will be written on SYS008. This tape
record will contain statistical information regarding the file as
well as the terms and their document numbers. In addition, all de-
scriptors which are input to the inverted file will reside on this
tape.
2.
Disk
The inverted file itself will reside on a disk. It will be com-
posed only of precise and temporary descriptors and their associated
document numbers. The disk will be created on SYS0l4.
-96-
-------
INVERTED FILE PRINT PROGRAM (IFPRNT)
Purpose
For maintenance purposes it will occasionally
complete listing of all of the members of the
gram furnishes a formatted dump of the entire
be necessary to obtain a
inverted file. This pro-
file on tape.
Input
The inverted file tape will be mounted on SYS009.
Processing
IFPRNT reads the inverted file tape and prints each precise or temporary
descriptor with its associated internal document numbers. When a common
descriptor is encountered, a message is printed below the term indicating
the total number of documents in which it occurs. Special descriptors
(unverified descriptors) are printed at the end of the listing and are
preceded by a dollar sign ($). The document numbers with which these
special descriptors are associated are also listed. When the end of the
inverted tape file is reached, processing terminates.
Output
The file will be listed on the printer.
is shown below.
A sample of the output listing
-97-
-------
FDA - J N ~_~R TED
F I L E
LISTING
PAGE
1
,.----- --
ACETYLOIGITOXIN
22,- __31 - - ------ ------ --
-~--~-~-
------- --
A~fTYt.~AUDJ..._~~CID
14, 15, 17, 42, 72. 76, 88, 107. 110, 111, 116. 159
ADRENAL CORTICOSTEROID
75
---- -------
-~ EN ERG I C _AG ENTS
1. 20. 110
AGITATION
42, 95, 96
ALCOHOL
129
AMBL YOP IA
82
1
__~!1INOPHYll INE
2, 143
AMINOSALICYLATE SODIUM
127
'\)
~
\
AMINOSALICYLIC ACID
7B, 80. 109. 111. 128. 170
AMOBARBITAL
90
AMOBARBITAL SODIUM
10
AMPICILLIN
16, 20, 25, 34. 54. 105, 106. 107, 1l0. 155. 168
AMPICILLIN TRIHYDRATE
139
AMYTAL
90
ANALGESIC-ANTIPYRETIC
14.15,11.42.12.16,88.107,110, Ill, 116.129.159
ANAPHYl/CIRCULAT/FAIl
7, 167
.~---
-98-
-------
INVERTED FILE STATISTICS PRINT PROGRAM (IVPRINT)
Purpose
In order to gain a rapid status of the descriptors used to index documents,
an indication of the activity of each of these may be desirable. The stat-
istics print program will provide data such as when was the descriptor last
used and how many documents does it describe. This information may aid
the indexer in the use of the descriptors. For example, if a descriptor
is used in a great many documents, it may not have the precision required
for rapid retrieval. In this case, it may be necessary to change its
function to common instead of precise. .
Input
The inverted file tape will be mounted on SYS007. A control card which
has the desired header will be read on SYSRDR. All data on the card will
be printed.
Processing
IVPRINT processes the statistical data preceding the document strings
on the inverted file tape. The term is printed along with the date it
was last added to the master file as a descriptor, the date it was last
deleted from the master file, the total number of documents in which it
is used, and the function of the descriptor. Each term is processed
separately. Processing terminates when an end-of-file is encountered
on the input tape.
-99-
-------
Output
The printed statistics
tape should be used to
tains only a channell
shown below.
will be written on SYSLST. The special FDA carriage
ensure correct processing. This carriage tape con-
punch. An example of the type of output expected is
-100-
-------
-------- ._~~--
----__-!JII.t.fRTE~FIlE- STAn STICS PRINT PROGRAM
J!.I;s..tBIPTOR
~~HYLOIGITOXIN
COUNT
-~!:.~U,-_S~U~YUL~<::_IQ_---- ---------
_AOR E~AL_~Q_R T 1~9~1 ERJJ-lQ-
.------- ._~~--
2
12
I
3
3
1
1
2
1
6
1
1
11
1
1
13
2
2
1
2
52
2
14
1
2
55
3
4
-101-
----- -----~---~---._- --~---~~ ---
._-~-
06/11168
LAST ADOlf ION
06 (11f 68
--- RAG E -..L..--
FUNCTION
LAST DELET! D~
PRECISE
06/11/68
PRECI~E
PRECISE
..MRf_NJ;R~ILA~J;1!I.L-
~U_ATlON
-------~--~- -------
06/11/68
-'----pRECISE
PRECISE
~--
-----
06/11/68
~J.~QHOJ-
06/11168
PRECISE
~-~--_.-
--'------_._--------
06/11/68
_____pREC I SE
PRECISE
AM8LYOPIA
.MINOP HY!-L INJ_____-
_AMINOSA~ICYlATE SOD1UM
--~--------~----~- ---~---~------
06111/68
06/11/68
PRECISE
.~~_.-
06/11/68
P~ECISE
PRECISE
.---- - -- -
~'liNOS_AL Iq'-lU~JIL______--
_AM08ARB IIAL__- ----
- ---_._----~----------~~---
AM08ARBITAl SOOIUlL______---
~_MF'.!CIlI..IN
--~---- -------~--
-------
I
"
t:::.
,-
\
_~M-')I CI '-I.._INJR.LHYDRA II; - -
AMYTAl
PRECISE
---~ -- ---_._--~
-- -- ----~-----------------
_Ati~LGESJ(;=ANT IP_YR ~TJc-
-._----- ---------
..Ati!~I:i)'llC I~CU_LAI!t=.A1,-__--
~_!:MIA
06(11/68
06111/68
PREC I SE
06/11168-
_.Q~[!1/68
~~~CJ S E
_f)Rf_(ISE_-
---- -~ -------~.~--
06111/68
PRECISE
06/11/68
06/11/68
- _PR~CI~E
PRECISE
06/11/68
PRECISE
PRECISE
._-------
_A_NE..uLHEMOL Yn~JiSPD
~_NGIOEOEMA
~~T I::: IN FECT IVES
ANTI-I~FlAM~~TQR~_AGENT
06/11/68
06/11/68
PRECISE
06111168
PRECISE
06/11168
AN T ~_<:l!>- F ~ S1" _D_R 1,/(;1) -
2.~~_U_ARLHR !IJ_~::-...R HEUMAJ I C
06/11168
PRECISE
PRECISE
~ANTIBACTERIALS TOPICAL
,ANT!B lOTI CS
06/11/68
06111168
06/11/68
PRECISE
PRECISE
PRECISE
,_~tITL~QA_<.;!lI"A~ T 1)---
-ANTI CON V91....sAN U_--------------
06/11/68
PRECISE
06/11168
-------
DATA CELL DUMP PROGRAM
BLOCK CHANGE EDIT PHASE
(EDITBC)
Purpose
When a term is to be repl~~~~Qr Qel~t~d throughout the entire master file,
the modification reque~~is submitted to this phase. This phase will test
the validity of the requests and build a table of these modifications.
Then, during reorganization of the data cell, that table is used to update
the master file.
Input
EDITBC reads cards in the following format:
~lBCDELETE PHRENOLOGY
~2BCREPLACE ACUTE POLIOMYELITIS:POLIOMYELITIS
Only a 'replace' may be continued on the next card. The format of the
card is checked and a message printed if the format is in error. How-
ever, only the term which will be used as a replacement is checked for
validity (i.e., is compared to a valid term on the dictionary master
file). If a term to be altered is misspelled, it will either not be
found or the wrong term may be altered. Extreme care should be exercised
when using this option.
Processing
This phase reads an input card and checks the format for errors. If
none are found, the card type (replace or delete) is determined. If
the card is a delete, the term is moved to the output area, the delete
code is moved in, and the next card is read. In the case of a replace,
the term and its replacement are moved to the output area (with the
length of each) and the replace code is moved in. Since a replace card
may be continued, the next card is tested. If it is a continuation
card, the remainder of data is moved to the output area. If it is not
a continuation card, it is processed as above.
When an end-of-file occurs on the input device, the data in the output
area is sorted and passed on to the next phase.
-102-
-------
DATA CELL DUMP PROGRAM
DUMP PHASE (DCELLDMP)
Purpose
The purpose of this phase is to dump the master file onto magnetic tape.
Input
Input to DCELLDMP will be data cell records.
assigned to SYS031.
The data cell will be
Processing
This phase reads records from the data cell, performs bIQc~£hanges, and
writes out a blocked master file tape. Tapes of update information for
the inverted -~fi.le, cross reference file, and !D. file are also created
as a result of block changes.
Output
1.
Output will be a magentic tape file containing the entire master
file. The tape will be assigned to SYS007.
2.
Unsorted inverted file update will be assigned to SYS004.
3.
Unsorted cross reference update will be assigned to SYS005.
4.
!D.
file revisions will be assigned to SYS006.
~I03-
-------
DATA CELL RELOAD PROGRAM
(LOADCELL)
Purpose
When the tape of the updated master file has been created, the file is
reloaded onto the data cell by LOADCELL.
Input
The tape created by DCELLCMP is used as input to LOADCELL.
mounted on SYS~~4.
It will be
Processing
LOADCELL simply reads the tape records and writes them onto the data
cell, creating and updating an index to the master file during the
processing. The records are transferred to MPRINTI which acts as a sub.
routine to LOADCELL. Processing terminates when the end-of-file is
encountered.
Output
The records are written by LOADCELL onto the data cell. In addition,
there is an option to list the entire master file. MPRINTI is used
for this purpose. The printer format is described in the MPRINTI
documentation.
104-
-------
THE SEARCH SYSTEM
-/os-
-------
SEARCHING SYSTEM
The data files described in the preceding sections are maintained so that
information pertaining to a given topic may be extracted as needed, in
the form desired. The searching system is designed to perform this ex-
traction process.
Unlike the dictionary and master file maintenance systems, the search
system consists of a single program, not a series of programs. Given
a request for information from the files, the search program will first
validate that request to ensure that the request terms are, in fact,
valid ones. This is done by checking each term against the same dic-
tionary (on disk files) used to validate the terms used in the docu-
ments themselves.
If the request contains no errors, the actual searching process begins.
The descriptors (terms) of the request are first looked up in the in-
verted file, a process directly analogous to the searching of a printed
index. From the inverted file, the program constructs a list of docu-
ment numbers identifying those documents which are potential responses
to the request. This list is termed the 'candidate list'. The program
then inspects in detail every document identified on this candidate
list to determine whether or not the document does, in fact, match the
exact specifications of the request. If it does, that document record
is written out. If not, the program examines the next member of the
list. It proceeds in this manner until the candidate list is exhausted.
At this point, the documents which satisfy the request (the 'hits)
have been printed, or are recorded on a tape, or both, depending upon
the options selected by the requestor. The records on tape are suit-
able for further processing by special purpose programs, such as a
report generator program to produce statistical summaries or other forms
of reports.
As is evident from the description above, the program processes each
request an an entity, completing all action on one request before
beginning work on another. Thus, while it is possible to 'batch'
requests by stacking them one behind the other in the card reader,
the program treats each as a 'batch of one '. This design method was
adopted to permit modification of the manner in which requests are
submitted to accomodate remote terminals at some future time, if de-
sired.
-106-
-------
SEARCH
SYSTE M
SEARCH
REQUEST
ERROR
LISTING
EDIT
PHASE
LlSTI NG OF
DOCUMENTS
REQUESTED
-107-
DIRECT
ACCESS
01 CTIONARY
DIRECT AC-
CESS IN-
VERTED
FI LE
DIRECT
ACCESS
CROSS RE-
FERENCE FILE
-------
SEARCH PROGRAM INPUT SPECIFICATIONS
Introduction
This section describes the preparation of inquiries, or requests, for
the data files. It includes a description of the card formats used,
the options available, and the type of output available.
Input Formats
A single inquiry or request to the system is prepared in punched card
form. The general pattern of the cards is shown below. Columns 1 and
2 are reserved for identification of the particular request. Columns
3 and 4 provide a sequence number for cards within a request. Column 5
is the beginning of the word which identifies the type of card; a colon
is used to mark the end of the identifying word. The remainder of the
card is punched with appropriate data as described below. There are
six different types, some required and some optional.
1. (:
-------
a second card, beginning at column 12 or later, if desired.
must be a minimum of one, and a maximum of two title cards.
example of a pair of title cards is shown below.
There
An
0l.0lTITLE:
0102TITLE:
J.F. JACKSON, PHARMACOLOGY DEPT. RM 324, EXT 7741
SEARCH ON CHLOROPROMAZINE DERIVATIVES.
2.
Search Card
This card is used to supply various overall parameters to the
searching program. It contains information specifying which
TYPE of search, which FILE(s) are to be searched, which DATA
BASE(s) are to be searched, and the MAXIMUM number of responses
desired. Each of these various parameters is described below.
a.
TYPE of Search
There are three possible kinds of searches: document, Boolean,
and mixed. A document (document number) search is one in
which only certain specified documents, as identified by
their internal document numbers, are retrieved. That is,
given a list of document numbers, the system will retrieve
those records and print them out. A Boolean search is one
in which documen~ are retrieved if they contain some speci-
fied combination of descriptors, subdescriptors, and the
like. A mixed search is a combination of the above two
(i.e., certain specified documents, as identified by their
document numbers, will be retrieved if and only if the rec-
ords for those documents contain the specified descriptors
and/or subdescriptors).
b.
File to be Searched
The word 'file' in that instance refers to the physical
computer files on which the data is kept. There are three
such files: the master file, which contains the full record
for every document in the system including free text, the
inverted file, which is an index to the master file, and
the 10. file, which is a condensed version of the master
file. The 10. file contains only some of the descriptors
relating to each document, and contains no free text. The
choice of which file to search depends upon the aims of the
searcher. The 10. file search can be used whenever the
searcher wants to get an overview of the file contents using
very broad (common) descriptors. .The inverted file search
could be used as a 'first pass' at some topic to gain some
idea of how much data there is in the file on that topic.
-109-
-------
Of these three files, the amount of information produced at
the end of the search and the amount of time required for the
search vary proportionately. This is illustrated below.
KIND OF SEARCH
AMOUNT OF INFORMATION
SPEED
Master File
Maximum: all searchable
data, all free text, as
desired
Slowest
ID. File
Title, document number
part of searchable
section, no text
Moderately
Fast
Inverted File
Document number only
Very Fast
c.
Data Base(s) Searched
As mentioned above, the word 'file' in this search card refers
to the physical organization of the data. Each logically dis-
tinct group of information may be identified as a separate
data base (e.g., 'Human Adverse Reactions', 'Veterinary data',
'Legal Hearings', etc.). The user may specify that either all,
or only certain specified ones of these data bases be searched.
d.
Maximum Number of Responses
The system contains a built-in upper limit on the number of
responses to a given question. However, the requestor can
override this limit by specifying a number of his own choice.
The system will accept this specified number in lieu of the
internal limit. The specified number may be either larger or
smaller than the system limit.
All of the options in the search card are shown below. For each of
the four parameters there is a preferred (automatic) option which
the system will use if nothing is specified. That preferred option
is underlined below.
-110-
-------
BOOLEAN ~. ID FI LE ~ ALL
SEARCH: TYPE: MIXED; FILE' fv1ASTERfDATA BASE' HUM; MAX . system limit
DOCUMENT INVERTED VET~: user number
It is therefore acceptable to use a search card with nothing punched
in it beyond the word SEARCH:. Such a card would automatically result
in a Boolean search of the master file, searching all data bases and
producing up to the system limit of responses.
*
Multiple data bases may be specified (e.g., DATA BASE: HUM, VET,
LGL; is permitted
3.
This card informs the system of how much data the requestor wishes to
have printed out once a document which satisfies the requirements of
the search logic has been found. There are two print parameters, one
controlling the printing of the searchable section and the other
controlling the printing of the free text segments. The searchable
section contains the actual terms used for search purposes, while
the free text, which can extend over a number of segments or para-
graphs, carries only descriptive information not used in the search
process itself. The requestor may choose to print, or not, the
terms in the searchable section, and may choose to have all, none,
or selected segments of the free text printed out. This is illus-
trated below.
PRINT' TERMS' NO; TEXT
YES
ALL
. NONE
S 1" S 2 " S 3" . . . (up
to a maximum of ten)
Again, the underlining indicates the preferred option, which will
automatically be chosen by the program if nothing is specified.
Note that if the requestor does not specify any options, thus per-
mitting the program to choose NO and NONE, the only result of the
search, even though the master file is searched, will be the list
of document numbers satisfying the request. It should also be
noted that it is illogical, and therefore erroneous, to specify
any text with an inverted file search, since there is no text
available in that file. It is also invalid to specify available
for printing from that file.
-111-
-------
4.
Report Card
This card is used only if the requestor whishes to extract data
from the files for further computer processing. This is known as
the 'RPG Option', which permits the user to not only have the
results of his searches printed for visual inspection, but also
to have the extracted data written onto magnetic tape for
use in preparing special reports of various kinds. If the user
does not wish to prepare these reports, the report card and the
select card described below may be omitted.
The report card specifies the general nature of the information
to be selected for the report tape. There are two parameters to
be given. The first is REQID (requestor identification). This
consists of any four character set of symbols the user desires.
Its function is to uniquely identify this particular ~group of
selected information so that it may be identified by' the programs
which are to follow. The second parameter is OUTPUT, which
identifies the kind of information which the user wants selected.
There are four possibilities: descriptors, subdescriptors, values,
and text. An example of the report card follows:
\ r_-----\
REQID : 1234; OUTPUT: DESC~!, ~lliB-, t~::r;1
This card indicates to the program that, from the records which
satisfy the requirements of the search request, the user desires
a tape file which contains records for every descriptor, for every
subdescriptor, and for all free text segments of those document
records.
REPORT:
5.
Select Card
This card is used to further specify the nature of the data de-
sired for the RPG Option. It permits the user to select de-
scriptors, subdescriptors, and values associated with only certain
file/context codes. It also allows him to specify which free text
segments he wants. An example is given below.
SELECT: DESCR : F/CJ F/C~ F/C~...TEXT : S~~S2~...
(F/C represents desired file/context codes) '(maximum of ten text segments)
This select card will result in the extraction
scriptors, subdescriptors, or values which are
specified file/con\ext codes. Similarly, only
segments will be ex~racted. The entire record
by specifying ALL following both DESCR: and
of only those de-
associated with the
the specified text
will be selected
TEXT: above.
-112-
-------
6.
Request Card
This card, or group of cards, is used to hold the actual search
terms for the Boolean, document, or mixed search. The detailed
explanation of these request cards is given in the 'Request
Language Description'.
-113-
-------
SEARCH LANGUAGE DESCRIPTION
Since the PL/l character set is the basic standard in this system, the
following are acceptable in descriptors and sub descriptors:
A through Z; 0 through 9; ) (
following characters are used
operators:
. , * / % # @ ,,, = + therefore, the
in the search language as separators and
1.
&
logical AND
2.
r
logical NOT
3.
logical
OR
4.
colon (for numeric operator, EQ:)
5.
semi-colon (used as a comma)
6.
$
for truncation of descriptors
7.
>
right parentheses
8.
<
left parentheses
9.
?
end of request
The request language (the means of interrogating the file of indexed docu-
ments) is designed to provide a wide range of capabilities, from simple
'AND', 'OR', 'NOT' statements to complex statements involving sequences
and linkage of descriptors.
The fundamental abilities of the language permit the usual three Boolean
operators, 'AND', 'OR', and 'NOT'. The symbols '&', 'I', and ,-;, re-
spectively, will be used to represent these operators. A descriptor, as
interpreted by the search language, is that which appears between two
operators. That is, a descriptor, with its associated subdescriptors
and values is treated as a single complex descriptor. In the following
example,
A&B&C,
can represent a simple inquiry such as,
~ 114-
-------
(A) (B) (C)
GANATOL & RASH & RECOVERED
or can represent a more complex statement, such as,
(A) (B) (C)
GANATOL; RTE ADM ORAL; DAY DOSE GM; EQ . 2.0 & RASH; SEVERE & RECOVERED
In the first instance, the three elements of the questions separated by
'AND' operators are simple descriptors. In the second instance, de-
scriptor A is a complex one represented by the entire statement 'Gana-
tol; Rte Adm Oral; Day Dose Gm; EQ : 2.0', which is itself composed of:
A descriptor
A subdescriptor
A second subdescriptor
A single value
Ganatol
Rte Adm Oral
Day Dose Gm
2.0
Descriptor B is represented by 'Rash; Severe', which is composed of:
A descriptor
A subdescriptor
Rash
Severe
Descriptor C is represented by the simple descriptor 'Recovered', as
in the first example.
Unless specifically stated otherwise, the term 'descriptor' will be taken
to mean either a simple or a complex descriptor in the remainder of this
document. Rules for the formulation of complex descriptors are presented
later.
Descriptors may be combined in a large number of ways. The simplest
of these ways, a simple 'ANDing', has already been illustrated, as
A&B&C. The 'OR' is used in a similar manner, AlBic. The former states
that all three descriptors, A, B, and C, must appear on a document in
order for that document to satisfy the request (be a 'hit'). The latter
states that if anyone of the three descriptors appears in a document,
the document is hit.
Parentheses (less than and greater than signs) may be used freely to alter
the meaning, but must be used with care. Parentheses cause the entire ex-
pression enclosed within them to be evaluated as an entity, the result of
that evaluation then being considered as a sort of 'super-descriptor',
which is then analyzed with the remainder of the request. For example,
-115-
-------
consider the following:
l.
2.
3.
AIB&C
&C
A I
The first statement defines a hit as any document which contains either
A or both Band C.
The second defines a hit as one which contains either A or Band C.
The third defines a hit as one which contains either A or both Band C
,
which is the same as (1). This is an example of redundant parentheses,
which will not affect the search logic.
Nested parentheses are permissible, as in the following:
A&!E>
This decomposes into A&B&D or A&E. However, if the parentheses are
modified the same expression decomposes to different meanings.
Removing the outer parentheses givesA&B&IE , which decomposes into
A&B&C or A&B&D or E.
Removing the inner parentheses instead gives A& , which decomposes
into A&B&C or A&D or A&E.
Removing both sets gives A&B&CIDIE , which decomposes into A&B&C or D or
E.
Logically, there is no limit to the number of nested parentheses which
are possible, since the logical system deals with them one at a time.
However, other limitations impose a maximum of eight nested sets. This
does not imply that only eight sets may be used in an inquiry, only that
a maximum of eight nested sets are permitted.
The 'NOT' operator (-;) should be treated with special care since its
function is different from either & or I. AND (&) and OR (I) are binary
operators which imply a relationship between two descriptors. 'NOT' is
unary. It affects only the descriptor following the -,. 'NOT' implies
'AND', i.e., A-,B = A and not B. A&(BI-,C) is not permitted and must be
wri tten as A&B I AI C.
116-
-------
Two special capabilities of logical manipulation of descriptors are al-
lowed, in addition to the elementary Boolean operators. These are:
linkage and sequence.
1.
Linkage
Links perform a special qualifying function for descriptors and
subdescriptors. To specify that descriptors are to be linked
in a search request, the descriptors in question are to be en-
closed in parentheses preceded by the letter 'L' as below:
The 'AND' operator (&) must appear between descriptors. Neither
logical 'OR' (I) nor logical 'NOT' (-d is valid. Furthermore,
no further parentheses may appear within the parentheses enclosing
the linked descriptors (i.e., no nesting). If links are used,
the document will be a hit only if the descriptors appear in the
record and if they share at least one link in common.
Also partial linking of descriptors can be requested such that
the first two descriptors must share at least one link in com-
mon, the third descriptor must be linked with the first or second
descriptor or both, the fourth with the first, second or third,
and so on. To request this type of linkage, precede the de-
scriptors within parentheses with 'L2'. This is called 'ACCUMU-
LATIVE' linking.
There are three additional linking options which involve linking
of subdescriptors. The 'L12' option combines the links of those
subdescriptors specified in the input with their associated de-
scriptors before testing if the descriptors are linked together.
to if 'LI2' is requested, the document will be a hit only if the
descriptors and subdescriptors appear in the record and if each
descriptor/subdescriptor set is linked with each other such set
within the parentheses. The option 'LII' will result in hits
only if those documents contain the specified descriptors and
subdescriptors and if the descriptors share a common link and if
the subdescriptors also share a common link. The third option
for linking subdescriptors is 'L22'. When using this option, the
links of each subdescriptor are 'added to' the links of the as-
sociated descriptor similar to option 'LI2' above. Then the 'ACCU-
MULATIVE' linking is used. This means that for each hit the first
descriptor/subdescriptors set of links must match the second set in
at least one link, the third with the first or second set, and so
on.
-117-
-------
2.
Sequence
It is possible to specify that descriptors appear in a record in a
certain sequence. That is, one may require that descriptors A, B,
C, and D appear in a record in that sequence. Several possibilities
are apparent, as can be seen from the illustration below.
1.
---A---B---C---D---
descriptors in record, in sequence with
intervening descriptors (non-contiguous)
2.
---ABCD------------
descriptors in record, in sequence, and
are contiguous
3.
---D---C---B---A---
descriptors in reverse sequence, non-
contiguous
4.
------DCBA---------
descriptors in reverse sequence, con-
tiguous
This request language recognizes four ways of specifying sequential
searches. They are:
TYPE
RETRIEVES
1.
FC (forward contiguous)
Case 2 only
2.
FN (forward noncontiguous)
Case 1 and Case 2
3.
EC (either forward or reverse
contiguous)
Case 2 and Case 4
4.
EN (either forward or reverse
noncontiguous)
All Cases
A sequential search is specified by enclosing the
in parentheses, preceded by one of the four codes
FN, EC, or EN) as below.
desired descriptors
shown above (FC,
EN
As with link searches, the AND operator must appear between de-
scriptors, and no other operator is valid. No other parentheses
may appear within those enclosing the sequential search descriptors.
-118-
-------
Simple and Complex Descriptors
A simple descriptor is what the name implies, a descriptor alone, without
qualification by either subdescriptors or numeric values.
A complex descriptor is one which is qualified, or modified in some way,~
by either subdescriptors or numeric values, or both. Some possible pat-
terms of complex descriptors which may appear in a document are shown
below:
( SD= subdescriptor and
V= numeric value)
1.
DESCR A; SDA; SDB; SDC;...SDN
2.
DESCR A; VI; V2; V3; ...VN
3.
DESCR A; SDA; VI; SDB; SDC; V2; SDX...
4.
DESCR A; VI; V2; V3; V4; SDA; SDB; V5; V6; SDC; SDD; V7; V8; V9...
Subdescriptors always are associated with the preceding descriptor.
Numeric values always apply to the immediately preceding subdescriptor,
if any. If there is no subdescriptor, the value applies to the descriptor,
Note that this requires that a numeric value which does pertain to a de-
scriptor must follow immediately after that descriptor, and precede any
subdescriptors which may also pertain to that same descriptor. In example
four above, for example, values VI, V2, V3, and V4 apply to descriptor A,
while values V5 and V6 apply to subdescriptor B, and values V7, V8, and
V9 apply to subdescriptor D.
119-
-------
Structure of DescriptorjSubdescriptor Patterns
Inquiries involving descriptors and subdescriptors in combination are
written as shown below.
DESCRIPTOR; SUBDESCRIPTOR; SUBDESCRIPTOR; SUBDESCRIPTOR ...
As many subdescriptors as desired may follow a descriptor. The descriptor
is written first. Each subdescriptor must be preceded by a semi-colon.
Thus, there is no semi-colon following the last subdescriptor. Spaces
before, after, or on both sides of the semi-colon are permitted but not
required.
All subdescriptors following a descriptor in an inquiry must be present
in the document record if the document is to be a 'hit'. This, in effect,
means that an 'AND' relationship is implied between a descriptor and its
associated subdescriptor(s). No parentheses may be used in a subdescriptor
string.
One may not phrase an inquiry such as 'DESCRIPTOR A with SUBDESCRIPTOR B
or SUBDESCRIPTOR C' as a single complex expression. That is,
DESCR A;
is not permissible.
The question must be written as:
DESCR A; SDB I DESCR Aj SDC
In phrasing subdescriptor questions, it will be convenient to treat the
semi-colon preceding a subdescriptor as meaning 'with' (or 'and with').
Thus, for example, the question:
A; SDX; SDY & B; SDZ
Would read as:
'Descriptor A with subdescriptor X and with subdescriptor Y and descriptor B
with subdescriptor Z'
-120-
-------
Truncation
It is possible to make inquiries of the system using only partial descriptors
by use of the truncation feature. In the normal, or untruncated, mode a
descriptor in the inquiry must match exactly with a descriptor in the rec-
ord order for the record to be a hit. Using the truncation feature, it is
possible to ignore terminal characters of descriptors in the record. In
the following inquiry, for example,
ROADS&ACC I DE~jTS
the descriptor 'ROADS' and the descriptor 'ACCIDENTS' must appear exactly
as written in the document in order to be a hit. In,
ROAD$&ACCIDENT$
however, any descriptor such as ROAD, ROADS, ROADWAY, and the like, would
match ROAD$. Similarly, ACCIDENT, ACCIDENTS, and ACCIDENTAL would match
ACCIDENT$. Truncation is thus a useful feature for avoiding the necessity
of 'ORing' together singular, plural, adjectival forms of the same word
or words. The program will ignore any characters appearing in the docu-
ment descriptors beyond those given in the inquiry. The truncation symbol
(the $) acts to block further comparison.
Truncation should be used with care, however, as unexpected retrieval might
result from its injudicious use. For instance, in the above example, ROAD$
will match successfully against not only the terms given above, but also
against ROAD SIGNS, ROAD BUILDING, and any other descriptor, regardless of
length which begins with the four characters ROAD. Similarly, ACCIDENT
PREVENTION, ACCIDENTAL LOSS, and any other descriptor whose first eight
characters are ACCIDENT will match successfully against ACCIDENT$.
The truncation feature may be used for
The only constraint is that there must
ceding the $. An inquiry such as:
both descriptors and subdescriptors.
be a minimum of one character pre-
ROADS & $
is invalid and will be rejected by the system.
The exact position of the truncation symbol has an effect on the retrieval.
ACCIDENT$ does not have the same meaning as ACCIDENT $. In the former case
(no blank between the word and the $), the eight characters ACCIDENT are
compared against the document records as described above. In the latter
case, the nine characters A-C-C-I-D-E-N-T-blank are compared against the
record. In this case, only such descriptors as ACCIDENT PREVENTION, and
other multiple word descriptors beginning with the word ACCIDENT would be
retrieved. Neither ACCIDENTS nor ACCIDENTAL would be retrieved. Further-
more, not even the simple descriptor ACCIDENT would be retrieved, because
-121-
-------
~escriptors are assumed to end with the last significant character. That
1S, they do not have trailing blanks. Therefore, there would be no match
between the eight character ACCIDENT in the record and the nine character
,
ACCIDENT $ in the inquiry.
Structure of Descriptor/Numeric Value Patterns
Descriptors qualified by numeric values are written using the operators
shown below. Each operator precedes its numeric value. Note that the
colon following each numeric operator is part of that operator and
must be included:
1. EQ: Equal
2. NE: Not equal
3. GE: Greater than or equal
4. GT: Greater than
5. LE: Less than or equal
6. LT: Less than
A descriptor with a numeric value is written as below:
DESCR A; EQ: XXX
TIle numeric operator (EQ:) and the value itself (here represented as xxx)
must be preceded by a semi colon. The numeric value may contain up to
nine significant digits, and is written in the conventional form. If no
decimal point is given, it is assumed to follow the last digit of the
value.
The operators EQ: and NE: may never be used with any other operator. GE:,
FT:, LE:, and LT: may be paired with one of the opposite type or may be
used alone. All allowable combinations of numeric operators are shown
on the next page.
-122-
-------
1. EQ: XXX
2. NE: XXX
3. GE: XXX
4. GT: XXX
5. LE: XXX
6. LT: XXX
7. GE: XXX LT: YYY or LT: YYY GE: XXX XXX must be less than YYY
8. GE: XXX LE: YYY or LE: YYY GE: XXX XXX must be less then YYY
9. GT: XXX LT: YYY or LT: YYY GT: XXX XXX must be less than YYY
10. GT: XXX LE: YYY or LE: YYY GT: XXX XXX must be less than YYY
Note that when paired numeric values are used, as in GE: XXX LT: YYY,
there is no semi colon between the first numeric value and the operator
preceding the second value. The lack of a semi colon here means that a
single value, which falls within the desired range from GE: XXX TO LE:
YYY is desired. A semi colon between GE: XXX; LT: YYY means that two
values are desired, one of which is GE: XXX; the other of which is
LT: YYY.
Numeric Value Precision
The precision of the numeric values used (i.e., the number of decimal
points included) will affect the search results of using either the
NE: or the EQ: operators.
When using GE:, GT:, LE:, or LT:, the numeric value given in the inquiry
is extended to the system maximum of nine significant digits by adding
trailing zeros as required. 2.15 becomes 2.15000000. Values in the
document record are compared against this extended number. However, for
EQ: and NE: the test for equality/inequality extends only to the level
of precision given in the inquiry. That is, in
DESCR A; EQ: 2.15
any value from 2.15000000 through 2.15999999 following descriptor A is
a hit.
123-
-------
Similarly, in
DESCR A; NE: 2.15
any value from 2.15000000 through 2.15999999 is a no hit.
Structure of DescriptorjSubdescriptor/Numeric Value Patterns
Both subdescriptors and numeric values may be used with descriptors as
long as the appropriate rules are observed.
DESCR A; EQ: XXX; SDA; LE: YYY; SDB; SDC; GT: III LE: QQQ
is a valid statement for example.
-124-
-------
SEARCHING SYSTEM
Purpose
This program is designed to accept requests from users, to validate those
requests against the dictionary, to search the data files of the system,
and to produce copies of those records in the file which meet the para-
meters of the search request.
Input
The input consists of request cards (outlined in search language de-
scription) and the files to be searched.
1.
Tape
The identification file tape is mounted on SYS007.
2.
Disk
a.
The cross reference disk file is mounted on SYS0l8.
b.
The inverted file disk is mounted on SYS0l4.
c.
The dictionary disk file is mounted on SYS0l0.
3.
Data Cell
The master file data cell is assigned to SYS03l.
Processing
This program operates in three logical phases. The first phase is the
editing of the input request, to ensure that all parameters are satis-
factory. Errors found in this validation operation will result in can-
cellation of the search attempt. For example, the use of the term in
the request which is not an authorized descriptor, as determined by a
-125-
-------
search of the dictionary, would render further processing futile, since
there is no such term in the file. Therefore, while the program will
continue to edit the remainder of the terms in the request (to detect
additional errors, if any), no searching will be performed. The pro-
cessing of that erroneous request will be terminated and appropriate
messages printed.
If the request is valid, however, the program proceeds to the actual
searching. The usual form of search is the combined file method,
making use of the inverted file on disk and the master file on the
large capacity direct access device.
The second phase of the program, in this usual mode of operation, con-
sists of a search of the inverted file which is an index to the master
file. By examining the strings of document numbers following the de-
scriptors in the request, the program is able to produce a 'candidate
list' of potentially responsive documents. This list represents the
output of this second phase.
That list is passed on to the third phase, the searching of the master
file itself. Each document in that list is retrieved from the storage
device and is examined in detail to determine whether or not it does,
in fact, satisfy all parameters of the request. If it does, it is
printed, written on tape, or both, as directed by the requestor. If
it does not, it is ignored and the program examines the next member of
the list. This process continues until the list is exhausted.
As described in 'search language description', there are several vari-
ations of this searching method. One such variation is an inverted
file search only. In this method, the search is terminated at the end
of the inverted file, producing only the candidate list in the form of
a list of document numbers.
Another variation is the ID. file search. The ID. file is a subset
of the master file containing only certain selected descriptors and
no free text. If a search involving very common terms is desired, it
may be preferable to search this ID. file since in certain circum-
stances it will be faster than a search of the master file.
It is also possible to search the tape, rather than the direct access
version of the master file itself. This form of search is available
for two reasons. First, it offers a back-up method of searching if
the direct access device becomes unavailable for any reason. Second,
FDA may wish to search its files at some computer installation which
does not have the necessary storage devices.
-126-
-------
Minor variations of the basic search technique included the 'document
number search' which is simply a request to extract a given document
or group of documents based upon their document numbers, and the 'mixed
search' mode which is a combination of the document number search with
the customary logical search. Both are explained in more detail in
the 'search language description'.
Output
Output is of two general kinds. If a request contains errors which
make a successful search impossible or highly improbable, messages
to this effect are printed, and the search for that request terminated.
If there are no errors, the output consists of either or both of the
following:
1.
A printed listing of documents which satisfy the parameters of
the request. A variable amount of information may be printed,
depending upon the requestor's specifications.
2.
A tape record of some or all of the searchable section of docu-
ments which satisfy the request parameters. It is suitable for
further processing via report programs or other special purpose
programs which may be written. This tape must be mounted on
SYS003.
A scratch disk must be mounted on SYS020.
file document number strings.
It is used for inverted
-127-
------- |