DOCUMENT RESUME ED 034 555 - ERIC · ed 034 555. authop. tttt.e. institution pub date. note....
84
ED 034 555 AUTHOP TTTT.E INSTITUTION PUB DATE NOTE AVAILABLE FROM EDRS PRICE DESCRIPTORS IDENTIFIERS ABSTRACT DOCUMENT RESUME LI 001 661 Austin, Charles J. MEDLARS, 1963-1967. National Library of Medicine, Bethesda, Md. f 68] 83p. Superintendent of Documents, Government Printing Office, Washington, D.C. 20402 (FS 2.202:M46/3, $0.75) EDRS Price MF-$0.50 HC Not Available from EDRS. Book Catalogs, Computer Programs, *Computers, Indexing, *Information Retrieval, *Information Storage, *Information Systems, Publications, Search Strategies, Subject Index Terms *Medical Literature Analysis and Retrieval System, MEDLARS, National Library of Medicine The purpose of this document is to present a final description of the original MEDLARS system as it evolved through four years of operation. The system is described as it was functioning on January 1, 1968. Among the various system elements discussed are: (1) the input subsystem, including journal selection and coverage, Medical Subject Headings (MeSH) , and indexing; (2) the retrieval subsystem, including request analysis, search formulation, file search, and printout of retrieved citations; and (3) the publication subsystem, including the MEDLARS photocomposer and computer programs for producing MEDLARS publications. A summary of operating experience for the .period includes discussion of system problems, changes, and evaluation. Also discussed is the extension of MEDLARS technology to the cataloging of books and serial titles. This system produces two major products: (1) catalog cards for the central NLM card catalog and (2) the NLM current catalog, a computer-produced book-form catalog available to other medical libraries on a current and frequent basis for use as an acquisitions and cataloging tool. Appended is a list of approximately 200 selected references. (JW)
Transcript of DOCUMENT RESUME ED 034 555 - ERIC · ed 034 555. authop. tttt.e. institution pub date. note....
ED 034 555
AUTHOPTTTT.E
INSTITUTIONPUB DATENOTEAVAILABLE FROM
EDRS PRICEDESCRIPTORS
IDENTIFIERS
ABSTRACT
DOCUMENT RESUME
LI 001 661
Austin, Charles J.MEDLARS, 1963-1967.National Library of Medicine, Bethesda, Md.
f 68]83p.Superintendent of Documents, Government PrintingOffice, Washington, D.C. 20402 (FS 2.202:M46/3,
$0.75)
EDRS Price MF-$0.50 HC Not Available from EDRS.Book Catalogs, Computer Programs, *Computers,Indexing, *Information Retrieval, *InformationStorage, *Information Systems, Publications, SearchStrategies, Subject Index Terms*Medical Literature Analysis and Retrieval System,MEDLARS, National Library of Medicine
The purpose of this document is to present a finaldescription of the original MEDLARS system as it evolved through four
years of operation. The system is described as it was functioning on
January 1, 1968. Among the various system elements discussed are: (1)
the input subsystem, including journal selection and coverage,Medical Subject Headings (MeSH) , and indexing; (2) the retrieval
subsystem, including request analysis, search formulation, filesearch, and printout of retrieved citations; and (3) the publicationsubsystem, including the MEDLARS photocomposer and computer programsfor producing MEDLARS publications. A summary of operating experiencefor the .period includes discussion of system problems, changes, andevaluation. Also discussed is the extension of MEDLARS technology tothe cataloging of books and serial titles. This system produces two
major products: (1) catalog cards for the central NLM card catalog
and (2) the NLM current catalog, a computer-produced book-formcatalog available to other medical libraries on a current andfrequent basis for use as an acquisitions and cataloging tool.Appended is a list of approximately 200 selected references. (JW)
IMO
En11.1
an
CZ104=1.=)
......?.
....'Q.
r, MNX 'WED
C.',CVCVCVCV
10.11
MN11111
MINI
0000
;", TrtO4,11 1..7 ....el. ......
CVCVCV
1.11
LLgnra)CC
II
.j
we41,
1
ççc1ioa3
zN
II co Ew
4 En 0
..1 g
FitW -J
z
ite
m g
%)9 tOO 4
*V10,
49 oLe
0-
r C
RES!I I Iliii I
0000000000011!23334353633333340414'
1 1 1 1 1 1 1 1 1 1 1
2 2 1 2 2 2 2 2 2 2 2
3 3 3 3 3 3 3 3 1 3 1
4 4 4 4 4 4 4 4 4 4
MEDLARS1963-1967
BY CHARLES J. AUSTIN
U.S. DEPARTMENT OF HEALTH, EDUCATION & WELFARE
OFFICE OF EDUCATION
THIS OOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE
PERSON OR ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINIONS
STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION
POSITION OR POLICY.
U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE
Public Health Service
National Institutes of Health
National Library of Medicine
Bethesda, Maryland 20014
ii
Mr. Charles J. Austin is Director of Computer Services and Assistant Professor ofPreventive Medicine, University of Colorado Medical Center, Denver, Colorado. He wasformerly Chief of the Information Systems Division at the National Library ofMedicine.
For sale by the Superintendent of Documents, U.S. Government Printing OfficeWashington, D.C. 20402 Price 75 cents
FOREWORD
At the National Library of Medicine, we regard the sharing of information with otherworkers in the field of information retrieval, documentation, and library science as one ofour most important responsibilities. In accordance with this philosophy, we are reportinghere the experience gained from more than f our years' operation of a large-scale referencestorage and retrieval system, MEDLARS.
I believe that Mr. Charles Austin's report describes our experiences, both good and bad,with objectivity. No attempt has been made to minimize the mistakes which were madeduring this experiment. There is a frank discussion of how well system objectives have beenmet.
The Library is currently in the process of designing a new system to replace MEDLARS.We earnestly hope that this final description of our first computer-based system will behelpful to others entering this field. We hope also that the lessons we have learned from ac-tually operating MEDLARS will be valuable to us as we move ahead toward applying moreadvanced techniques of information processing. The documentation of the experiences ofothers, added to our own, should provide information of great value to the managers as wellas the users of computer-based information storage and retrieval systems.
Martin M. Cummings, M.D.DirectorNational Library of Medicine
iii
CONTENTS
CONTENTS
LIST OF FIGURES.
LIST OF TABLES
DEFINITION OF FLOW CHARTING SYMBOLS
vi
vii
viii
Chapter I. INTRODUCTION 1
Background and History 1
System Objectives 3Design Criteria 4
II. SYSTEM DESCRIPTION 5
Major Products of the System 5
Overall Data Flow 5
Equipment 7
Staff 9
III. INPUT SUBSYSTEM 13
Coverage and Selection of Journals 13
Medical Subject Headings 13
Indexing and Revision 16
Input Preparation and Verification 20
Input Module 22
IV. RETRIEVAL SUBSYSTEM 25Request Analysis and Search Formulation 25Preparing the Search for Computer Input 29Searching the Computer Files (Demand
Search Module) 33Printing the Retrieved Citations (Report
Generator Module) 34
V. PUBLICATION SUBSYSTEM 37Establishing and Testing Recurring
Bibliographies 37Computer Programs for Producing
MEDLARS Publications 37
The MEDLARS Photocomposer(GRACE) 41
Final Processing 41
VI. UTILITY PROGRAMS 45Updating the Subject Headings File
(MeSH Generator Modulo) 45Maintaining the CCF File (File Main-
tenance Module) 45
PageKeeping Records of System Activity
(Statistical Module) 48Miscellaneous Utility Programs 48
VII. OPERATING EXPERIENCE: 1963 -1967. 51Data Conversion Period 51Problems with GRACE 51Initial Production Period 54Maintenance of the System; Changes
Made 54MEDLARS Decentralization 56Formal Evaluation of MEDLARS 58Performance in Relation to Original Sys-
tem Objectives 59
VIII. THE CURRENT CATALOGAN EXTEN-SION OF MEDLARS 61
Background and History 61products of the System 61Equipment Used 64Information Flow and Procedures 64Operating Experience 68
IX. SUMMARY 69SELECTED REFERENCES 71
List of FiguresFigure
1. Definition of Flow Charting SymbolsPage
vi2. Specimen Page of 1879 Index Medicus 23. Demand Bibliographies 84. MeSH Statistical Report 95. Flow ChartMEDLARS System .10
6. NLM Staff Working on MEDLARS 127. Flow ChartInput Subsystem 148. Stamped MEDLARS Journal 169. Specimen Pages from MeSH 17
10. Specimen Page MeSH Tree Structure Listing 1811. Indexer Data Form 1912. Paper Tape Strip and Flexowriter Hard Copy 2113. Paper Tape Reader 2214. CCF Magnetic Tape Record Layout _ 2315. Specimen MEDLARS Search Request 2616. Specimen Demand Search Formulation Record 2717. Specimen Report Generator Request Form 2918. Flow ChartRetrieval Subsystem 3019. Specimen Search Appraisal Form 35
vi
Figure Page20. Flow ChartPublication Subsystem 3821. Sample Index Medicus Page 4222. Schematic Diagram of GRACE Optical Unit 4323. Flow ChartMeSH Generator 4624. Flow ChartFile Maintenance Module 47
25. Flow ChartStatistical Module 4926. Specimen PageJanuary 1964 Index Medicus 5227. Specimen PageJuly 1964 Index Medicus 5328. Demand SearchAverage Processing Time 55
29. Demand Searches Completed 56
30. Specimen PageNLM Current Catalog 62
31. Specimen Catalog Cards Produced by Computer 63
32. Flow ChartComputer-Aided Catalog System 65
33. Specimen Catalog Master (Input Data Sheet) 67
List of Tables
Table No. Page1. MEDLARS Publications 6
MEDLARS Data Processing Equipment2. 11
3. Changes in the Medical Subject Headings File 16
4, MEDLARS Search Elements 28
Decision Table Logic5. 34
6. Statistical Module Reports 48
7. Failure Analysis Table 59
8. Operating Statistics: 1966-1967 68
VII
Paper documents and reports ofall varieties.
\MANUAL A manual offline operation not re-
OPERA- quiring mechanized aid.
TION
COMPUTER
OPERATION
viii
An operation utilizing a key-drivendevice.
A major processing function ac-complished by computer.
Magnetic Tape
Punched Paper Tape
Punched Card(s)
Yes
Points in the system where differ-ent paths or branches are fol-lowed based upon the results ofa yes-no decision.
Offline storage of dataprintedreports, cards, magnetic or papertape.
Switch. Path through system whichis followed only when certain spe-
'1.°°...1." cial conditions are met.
FIGURE 1.Definition of Flow Charting Symbols.
Connectors. An entry from, or anexit to, another part of the flowchart.
Chapter I
INTRODUCTION
In January 1964, after three years of de-tailed planning and system development, theNational Library of Medicine initiated MED-LARS (Medical Literature Analysis and Re-trieval System). The original system designwas described in detail in The MEDLARSStory* published in 1963. Various referencesto the system as it developed can be foundin the bibliography of this report.
The purpose of this document is to presenta final description of the system as it hasevolved through a period of four years ofoperation. This will add the final chapter tothe present MEIN story at a time whenthe Library is on the threshold of developingan entirely new system, utilizing the latesttechniques of documentation and informa-tion science coupled with modern, "third-generation" computer equipment.
The reader interested in a complete chron-icle of the Library's experience with MED-LARS is advised to combine the reading ofthis report with a re-reading of the originalMEDLARS story. Thus he will be able todevelop a complete picture that answers thequestions: (1) What did the Library set outto do? (2) What was actually accomplished?and (3) What changes were made in theoriginal system design and why?
MEDLARS is a dynamic system which isconstantly changing; therefore, any systemdescription must necessarily reflect only asnapshot view of a rapid motion picture.This report describes the system as it wasfunctioning on January 1, 1968. Althoughsome additional changes will be made before
*The MEDLARS Story at the National Library ofMedicine, U.S. Dept. of Health, Education, andWelfare, Public Health Service, Bethesda, Maryland,1963,
the new "third-generation" system goes intoproduction, this point in time is adequate todescribe, at least in general terms, the"final- version of the original MEDLARS.
In addition to describing the final pictureof MEDLARS as it evolved over the years,this report emphasizes the changes madeduring the evolutionary process. Problemsencountered and unmet objectives are franklydiscussed in recounting the MEDLARS ex-perience. The formal evaluation of MED-LARS conducted during 1966 and 1967 addsto the reservoir of data collected on systemperformance. It is hoped that this detailedreporting on the Library's experience will as-sist others involved in the design and opera-tion of information retrieval systems.
Background and HistoryThe information. which MEDLARS pro-
cesses and disseminates consists of referencesto published medical literature. Effectiveutilization of published literature requiresbibliographic control through descriptive andsubject indexing of individual books andjournal articles.
The Library initiated its program of bib-liographic control of the medical literaturein 1879 with publication of the first issue ofIndex Medicus, which continued until 1927(see Figure 2). Replaced from 1927 to 1956by the Quarterly Cumulative Index Medicus,published by the American Medical Associa-tion, Index Medicus reappeared as a Librarypublication in 1960, replacing the monthlyCurrent List of Medical Literature.
From 1960 to 1963, Index Medicus wasproduced by a partially mechanized systemthat utilized paper tape typewriters, punched
1
2
Index Medicus. [No. 1. Jan. 3e, '79.
ingulo interno de los pirpados y queratitisurativa en el ojo derecho de un recien nacido;accidentes sifiliticos generales consecutivascuracion. Rev. de med. y dr. prfict., Madrid,1878, ii, 241-246. Chambard (E.) Gommcdu foie. Note additionnelle a la presentation fieM. Gille. Progres med., Par., 1878, vi, 764-765. Cheever (D. W.) Syphilitic constrictionof the pharynx. Boston M. & S. J., 1878,xcix, 649. Coyne (P.) Contribution a l' etudede la syphilis cerebrale. J. d. conn. med. prat.,Par., 1878, 2 s., ii, 302 -306; 821-325. De-patil. Pemphigus syphilitique. J. d. sages-femme., Par., 1878, vi, 138. Dowse (T.S.) Syphilitic epilepsy. Practitioner, Lond., 1878.xxi, 270-282. Griinfeld (J.) Die Ham des&hankers als Charakteristicurn der Initialformvon Syphilis. Allg. Wien. med. Ztg., 1878, xxiii,v17. Hyde (I N.) The nurse-maid and theaiother of the syphilitic child. Chicago M. J. &Exam., 1878, xxxvii, 452-463. Jullien (L.)Etude sur les varietes et les formes anormalesdu chancre simple. Progres med., Par., 1878,vi, 657 ; 733-735 ; 759 ; 782-784 ; 797. Keyes,(E. L.) Multiple chancre of the nipple. Arch.Derma., N. 'Y., 1878, iv, 126-127. Slabs(E.) Leber' Syphilis-Impfung bei 'I'hieren unddie Natur des syphilitischen Contagiums. Allg.Wien. med. Ztg., 1878, xxiii, 418, also Prag.med. Welinschr., 1878, iii, 409-411. L. (J.R.) Syphilitic notes. North Car. M. J.,Wilmington, 1878, ii, 241-244. Le Pileup.31ennorrhagie uretbrale ayant pour originel'inoculation de pus d'une glande de Bar-
tholin abcedee. Traitement par le sante' jaune;guerison complete le 28 jour. Ann. de dermat.et syph., Par., 1878, ix, 874 -379. Miner (J.F.)Iodide of potassium in syphilis. Buffalo M. &S. J., 1878, xviii, 173-177. Obtulowicz (F.)Przyczynek do leczenia ostrego zapalenia wiewie-rowego cewki moczowej. Przegl. lek., Krakow,1878, xvii, 505-507 ; 517-519. PosaeIt (K.)Syphiloma cerebri. Ann. d. stiidt. allg. Kran-kenh. zu Munchen, 1878,1,299. Potain. Fievresyphilitique. Gaz. d. hop., Par., 1878,1i, 961.Prolate (G.) Sifilide cerebrate. In his "Clip.dermo-sifilit. di Palermo," 1878, 116-130.Prognosi di previsione della sitilide. Ibid., 131-149. Ancora della cure mercuriale ipo-dermicadella sifilide. Ibid., 150-166. de Banti(L.) Etude sur le retrecissement syphilitique durectum. Rev. med. dc Toulouse, 1878, xii, 193-210 ; 244-264; 282-303. von Sigmund. DieReklame ftir Tayuya gegen Syphilis. Wien.med. Wchnschr., 1878, xxviii, 979-980. VanBuren (W. IL) On nervous syphilis. Med.News & -Libr., Phila., 1878, xxxvi, 189-194.Whistler (W. M.) Lectures on syphilis of thelarynx. Med. Times & Gaz.; Lond., 1878, ii,343; 372; 405; 535; 650.
PARASITES.See, also, Diseases of the Skin.
Goubert (E.) Des yen chez les enfants, et 41maladies vermineuses. Paris, 1876, Dion. 163pp. 18 °.
Atkinson (I. E.) The botanical relations oftrichophyton tonsurans. N. York M. J., 1878,
xxviii, 562-575. Bourel- Ronciere. De ?he-matozoaire nematolde de illumine et de son im-portance pathogenique d'optes les travaux ang-lais et bresiliens de ces dernieres entrees. Arch.de med. nay., Par., 1878, xxx, 113-134; 19.2-214. Hizon (C.) Slippery elm bark for tapeworm. Ohio M. Recorder, Columbus, 1878, iii,256. -- Mitchell, D. Trichiniasis in Otsego Town-ship, Allegan Co. Rep. 'Bd. Health, Mich.,Lansing, 1878, lxxii-lxxiii. Normand. Durole etiologique de l'anguillule dans la diarriteede Cochinchine. Arch. de med. nay., Par.,1878, xxx, 214-224. Bangstlli. Perforation del'intestin grele intact par des ascarides durant lavie du malade. Cong. period. internat. d. se.med. Compte-rendu, Gemme, 1878, 247-250.Scheuthauer (G.) Kiisig zerfallende Ilerde inder Leber eines 4 jahrigen Knaben, bewirkt durchSpulwiirmer der Lebergallenglinge. Jahrb. f.Kinderh., Leipz., 1878, n. F., xiii, 63-69. SilvaAranjo (A. J. P.) Is Maria immitis et la Mariasanguinolenta au Bresil. Lyon med., 1878, xxix,319-327; 363-369. Sonsino. Nouvellee re-cherches Jur les hematozoaires de l'homme enEgypte. Cong. period. Internat. d. se. med.Compte-rendn, Geneve, 1878. 651-653. Turn-bull (C. S.) Filaria in the eye. Med. & Surg.Reporter, .Phila., 1878, xxxix, 351-355. Vogt,C. La provenance des entozoaires de l'hcaume.Cong. period. internat. d. sc. med. Compte-ren-du, Geneve, 1878, 105-141.MEDICAL GEOGRAPHY, STATISTICS AND
CLIMATOLOGY.
.Anales de la (Akins meteorologica Argentina.Por su direction Benjamin A. Gould. Tomo t,Clime de Buenos Aires. Buenos Aires, 1878,E. Coni. 522 pp. 17 pl. 4 °.
Astxonomische, magnetische und meteorolo-gische Beobachtungen an der k. k. Sternwartezu Prag. Auf Offentliehe Kosten herausgege-ben von Carl Hornstein. 88 Jahrgang (1877).Prag, 1878. 56 pp. 4°.
Bird (S. P.) On the influence of the Austra-lasian climates on imported plithisis. Austral. M.J., Melbourne, 1878,. xxiii, 84-40. Day (H.)On ozone in relation to health. Med. Press &Circ., Lond., 1878, n. I. xxvi, 285-286.Lobed. Die Curorte der Riviera di ponente inihrem therapeutischen Werth° in Bezug auf Kli-matologie und Seebiider. Berl. klin. Wchnsclir,1878, xv, 499; 530; 617-620. Lente, F. I).The constitution of climate, with special refer-ence to the climate of Florida. Riclu»ond & Lou-isville M. J. Louisville, 1878, xxvi, 361-370.McParlin (T. A.) Notes on the history and cli-mate of New Mexico. Rep. Smithson.'Inst., 1878.:321-348. White (C. B.) Some effects of alti-tude and dryness upon disease. Ohio M. 1-corder, Columbus, 1878, iii, 241-249.
HOSPITAL REPORTS AND CLINICAL.MEDICINE.
Set, also, Military and Naval Medicine.
Annalen der stiidtischen allgemeinen Kranken-hiiuser zu Miinchen. Itir Verein mit den Aerz-ten dieser Anstalten lierausgegeben von Ziems-len. Bd. i. Miinchen, 1878, G. Himmer.659 pp. 9 pl. 8°.
FIGURE 2.Specimen page of 1879 Index Medicus produced by Dr. John Shaw Billings.
card equipment, and a Listomatic Camera.*Operation of the Listomatic System greatlyaided in the subsequent development ofMEDLARS, since: (1) the earlier systemprovided much background data used in thedesign of MEDLARS; (2) it offered a valu-able operating experience on which to basethe system design; and (3) it assisted in thedata conversion task for MEDLARS, sincesome bibliographic data had already beenput in maaine-readable form for the Listo-matic process.
Although the Listomatic Camera Systemworked effectively in the publication of IndexMedicus and related publications, it had lim-itations. Perhaps the most serious of thesewas the very limited information retrievalcapability. The large number of punchedcards accumulated for only one year's publi-cation precluded any effective informationretrieval using punched card equipment(about 1/2 million new cards each year).
Primarily because of these limitations ofthe Listomatic System, and also because ofthe rapidly growing size of Index Medicus,the Library decided to begin planning a newand more highly mechanized system. Usingboth. outside consultants and members of thestaff, a set of overall system requirementswas prepared in early 1961. Because of lim-ited technical capability of the staff, a deci-sion was made to use an outside contractorfor the detailed design and implementationof the new system. Bidding and evaluationof contractors' submissions occurred duringthe first six months of 1961, and a contractwas awarded to the General Electric Com-pany, Information Systems Operation, Be-thesda, Maryland.
Phase I of the contract (from mid-Augustto December 1961) resulted in a set of pre-liminary system specifications including anoverall system plan, computer programmingrequirements, and evaluation and selection ofmajor equipment required for the system.
*For an excellent description of the ListomaticSystem see: S. I. Taine, The National Library ofMedicine Index Mechanization Project, Bulletin ofthe Medical Library Ass., 49: No. 1, Pt. 2, (Jan.1961).
General Electric Co., Final Report of Phase I:Preliminary Design, Bethesda, Maryland, 1962.
Phase II and Phase III of the contract (Jan-uary 1962December 1963) included detaileddesign and pre-production activities. Themajor tasks were: (1) computer program-ming and testing; (2) procedure writing;(3) ordering and installation of equipment;(4) recruiting and staffing of personnel tooperate and maintain the system; (5) con-version and final system testing prior to pro-duction (scheduled for January 1964).
The conversion period ran from April toDecember 1963. After installation of thecomputer in March 1963, approximately45,000 journal article citations from the1963 Index Medicus were converted to mag-netic tape. During this same period new per-sonnel were added to the Library staff, andan extensive system test was carried out.Cut-over to the new system was accomplishedin January 1964, and it has been in opera-tion continuously since that date.
System ObjectivesThe major objectives for the MEDLARS
system as stated by Library management in1961 are listed below:
(1) Improve the quality of and enlarge(broaden the scope of) Index Medi-cus and at the same time reduce thetime required to prepare the monthlyedition for printing from 22 to 5working days.
(2) Make possible the production ofother compilations similar to IndexMedicus in form and content (but inmore specific medical subject areasand hence smaller in size).
(3) Make possible, for Index Medicus andother compilations, the inclusionof citations derived from othersources, as well as from journalarticles.
(4) Make possible the prompt (a maxi-mum of two days) and efficient serv-icing of requests for special bibli-ographies, on both a demand and arecurring- basis, regularly searchingup to five years of stored computerfiles.
(5) Increase the average depth of index-ing per article (number of descrip-
3
tive subject terms per article) by afactor of five, i.e., ten headingsversus two.
(6) Nearly double the number of articlesthat may be handled (indexed andentered into the computer) annuallyfrom 140,000 now to 250,000 in1969.
(7) Reduce the need for duplicative totalliterature screening. operations (atother libraries and information cen-ters).
(8) Keep statistics and perform analysesof its own operations to provide theinformation needed to monitor andimprove system effectiveness.
(9) Permit future expansion to incorpor-ate new and as yet not completelydefinedand hence secondary ob-j ectives.
Some of these objectives were quite real-istic and have been effectively accomplished.Others turned out to be difficult or not feas-ible at the time, and have met with onlylimited success. A more complete discussionof how objectives have been accomplished iscontained in Chapter VII, Operating Expe-rience.
Design CriteriaAlthough it is a generally accepted principle
of systems design that the designer shouldoperate with as few predetermined con-straints on his imagination as possible, none-theless, a large-scale, operational system suchas MEDLARS usually requires that some"ground rules" be established in advance.
Some of the major guiding principles onwhich the design of MEDLARS was basedare discussed below.
First, and perhaps most important, was adecision to continue to use human indexersfor assigning subject descriptors to the liter-ature for subsequent retrieval and publica-tion of references. Library staff and thecontractor both agreed that the state of theart of automatic indexing in 1961 was suchthat it was not feasible for MEDLARS. Asecond important and closely related decisionwas to continue to use a controlled vocabu-lary for indexing, rather than experimentwith natural language or other "open-ended"
4
indexing approaches, none of which had everbeen used in a large production system.
Another major decision was to index eacharticle only once, and use a single computerinput record both for publication in IndexMedicus and for retrieval purposes. This pol-icy, which was adopted for economic rea-sons, has not been without its problems inrequiring compromises in actual operationof the system.
Other important design criteria included:(1) A decision to train search specialists
for formulating retrieval requestsfor the computer, rather than allowcustomers of the system to attemptto formulate their own computersearch statements.
(2) A decision to use serial magnetic tapefiles for storing journal article ci-tations, rather than random accessdevices. This decision was reachedafter a careful analysis of compara-tive costs and retrieval efficiencybased upon equipment available in1961-62.
A decision to segment computer pro-grams into self-contained "modules"for ease of maintenance and systemchanges.
(4) A requirement that the system employa "high-quality" output device, su-perior to available computer print-ers, for preparation of copy forMEDLARS publications. This re-quirement led to an extensive studyof possible output devices, and asubsequent decision that the Librarymust develop a high-speed, high-quality photocomposition device.The decision was a truly momentousoneboth for its impact on MED-LARS and the photocomposition in-dustry in general !
A decision not to increase the amountof clerical work required of the pro-fessional indexer by using clericalpersonnel for proparation of thecomputer input record. It was de-cided also to design the system so asto use the computer for as muchcoding and editing of the input dataas poSsible.
(3)
(5)
t.
Chapter II
SYSTEM DESCRIPTION
Major Products of the SystemThe products of MEDLARS can be di-
vided into two major categories : (1) bibli-ographic publications designed for use by alarge group of people working in relatedfields; and (2) individual demand searchesof the literature tailored to the stated re-quirements of an individual or small group ofpeople working on the same project.
Publications produced by MEDLARS as ofJanuary 1, 1968 are listed in Table 1. Thefirst six publications are general referenceworks used in medical libraries throughoutthe world. The last nine are more special-ized recurring bibliographies produced in avariety of formats in cooperation with pro-fessional societies and other governmentagencies working in the specialty fields.These cooperating organizations ptovide as-sistance and guidance in setting up the bib-liography, and take responsibility for print-ing and distribution to workers in the spe-cialty field.
Demand searches are individually formu-lated to meet the specifications of the re-quester. The bibliographies produced as a re-sult of the search are printed either on3" x 5" cards or on continuous computertabulating paper. The content of these de-mand bibliographies ranges from a few arti-cles in a very specialized field, to a broadreview of the literature for someone lookingfor a comprehensive listing of everythingrelated to his field of interest. Some sampledemand bibliographies are shown in Figure3.
Professional staff of the Library some-times formulate demand searches for an in-dividual which, they believe, may be ofbroader interest to others. These bibliogra-
phies are reprinted in high-quality formatas "Literature Searches" and copies are sentto anyone upon request. Updated lists ofLiterature Searches are regularly printed inPublic Health Reports, Journal of the Amer-ican Medical Association, Journal of theAmerican Dental Association, and Drug Re-search Reports; announcements are also pub-lished in other major journals and the bib-liographies are widely circulated at profes-sional meetings. (A complete list of currentLiterature Searches is available from theLibrary's Office of Public Information.)
In addition to publications and demandsearches, MEDLARS also produces internalreports used by operating and managementpersonnel. These include such things as oper-ating statistics and print-outs of computerfiles. Figure 4 is a sample statistical reportshowing the frequency of usage of MedicalSubject Headings (MeSH).
Overall Data FlowThe data flow through MEDLARS can best
be understood by referring to the flow chartin Figure 5.
The system can be described functionallyas consisting of three major parts : an InputSubsystem, a Retrieval Subsystem, and aPublication Subsystem.
The Input Subsystem combines the intellec-tual talents of trained literature analystswith the processing and storage capabilitiesof the computer. New medical journals arechecked in and forwarded to the Index Sec-tion, where the analysts classify the subjectcontent of each article in the journals by as-signing appropriate descriptors from theLibrary's controlled list of terms (MedicalSubject Headings). The indexers are respon-
5
'r 1.MEDLARS Publications (as of January 1, 1968)
Name Description Frequency
Index Medicus Comprehensive listing of articles indexed Monthlymajor library reference tool. Subject &author listings; separate section listing re-view articles
CooperatingOrganization
Cumulatedindex Medicus
Cumulation of monthly issuessubject cross Annualreferences added; no review section; jour-nals indexed for Index Medicus are listed
Medical SubjectHeadings
Alphabetic and categorized list of subject Annualheadings used in MEDLARS indexing. Pub-lished as Part II of January issue of IndexMedic=
List of JournalsIndexed inIndex Medic=
Journals indexed for Index Medicus listed Annualby title abbreviation, full title, major sub-ject areas, & country of origin
MonthlyBibliography ofMedical Reviews
Listing of review articles in monthly Index MonthlyMedicus; subject and author sections. Cum-ulated annually in CIM.
NLM CurrentCatalog
List of books, serials, and reports cata- Biweekly withloged; subject and author listings quarterly
cumulations
RECURRING BIBLIOGRAPHIES:
CerebrovascularBibliography
Subject bibliography with separate subject Quarterlyand author indexes
National Instituteof NeurologicalDiseases & Blind-ness & Nat'l HeartInstitute
Fibrinolysis, Subject and author sections, reviews, sub- MonthlyThrombolysis, and ject and author indexesBlood Clotting
National HeartInstitute
Index ofRheumatology
Subject and author sections Monthly AmericanRheumatismAssociation
Index to DentalLiterature
Subject & author sections; some additional Quarterlyjournals not included in Index Medicus areindexed for this publication
American DentalAssociation
InternationalNursing Index
Subject & name sections; some additional Quarterlyjournals not included in Index Medicus areindexed for this publication
American Journalof NursingCompany
Artificial KidneyBibliography
Articles listed by broad subject area; bib- Quarterlyliography is still being tested prior to gen-eral publication and release
National Instituteof Arthritis &Metabolic Diseases
Medical EducationBibliography
Published as a regular section of the Jour- Monthlynal of Medical Education
Amer. Associationof Medical Colleges
(Continued)6
TABLE 1.continued.
Name Description FrequencyCooperatingOrganization
Bibliography ofEndocrinology
Subdivided into major subject categories Bimonthly National Institute
and author section with separate subject of Arthritis and
and author indexes Metabolic Diseases
Surgery of theHand
Subject and author sections Quarterly Amer. Society forSurgery of the Hand
sible also for translation of foreign titles,and a trained clerk transliterates titles ofarticles written in non-Latin alphabets.Journals with indexer data sheets attachedare forwarded to the Input Section, whereclerk-typists prepare punched paper tape in-put for the computer. The basic unit recordconsists of the article's title, author names,journal reference, and subject headings as-signed by the indexer. The punched papertape is accompanied by typewritten copy thatis sight-verified by a staff of proofreaders.Once a day, all corrected paper tapes arebatched and spliced together for entry intothe computer.
The input computer programs accept thepaper tape, edit the data extensively, andprepare the major data filethe CompressedCitation File (CCF). The Compressed Cita-tion File is a highly coded, time-sequentialstore of citations used for searching and re-trieval, both for individual demand searchesand for MEDLARS publications.
The Retrieval Subsystem begins with thereceipt of a request for a demand bibliog-raphy. Such requests are forwarded to a staffof search specialists who have had extensivetraining in both indexing and the logic of acomputer search. These specialists formu-late the request into a list of search param-eters linked in logical fashion by up to threesearch statements. The formulated searchrequests are punched into cards and batchedfor daily computer processing. The search andretrieval programs match a batch of searchquestions against every record in the Com-pressed Citation File. Citations retrieved areprinted in any one of a variety of outputformats by means of print programs. The
resulting lists of citations are referred to as"demand bibliographies."
The Publication Subsystem is concernedwith preparation of periodic indexes to cur-rent biomedical literature. In accordance witha publication schedule, search specificationcards are entered into the computer forbibliographies to be compiled. The searchand retrieval programs retrieve the appro-priate citations from the CCF. The publi-cation format programs perform a rathercomplicated task of page composition andprepare a magnetic tape file of one-line printrecords for a computer phototypesetter.Thistape is used to convert the digital informa-tion from magnetic tape to characters onphotographic film or paper. The exposedfilm or paper is developed by an automaticprocessor, inspected, cut into page-size sheets,and packaged for mailing to the printer. Theresulting typeset pages then can be used di-rectly for printing the final publication.
Chapters III, IV, and V describe the threemajor subsystems in greater detail.
In addition to the major data flow de-scribed above, the system also employs var-ious utility programs for updating of masterfiles and preparation of statistical reportsused for management control. These utilityprograms are described in Chapter VI.
EquipmentTable 2 shows the data processing and re-
lated equipment used in MEDLARS.Input equipment includes paper tape type-
writers for punching indexed citations, andpunched card equipment for preparing cardsused for demand searches, updating of mas-ter files, and initial entry of computer pro-grams.
Current computer equipment includes aHoneywell-800 computer used for main pro-
7
07085?LA PLACA
CN THE MECHANISM CF ThE CYTCPATHIC CHANCESPRCCLCEC IN LIALIAo. Ams.:Trm (CII rilMOce Cu
THE PCLLUSC 070852RFPCRT). KAYE JW
ARCH CES VIP PRCBLEMS IN THERAPY CF MOLLUSCLMCONTAGICSUM. CASE REPCRT.
AMNICN. *C1HUMAN (4),
8
ARCH GERM (CHICAG(
*ALCCHOL, PROPYL/HUNAN (4), MALECCNTAGIOSUM/ DRUGCCNTAGIOSUM/ SURGITHERAPEUTIC USE
LEVIN HL070852
BACTERICSTASIS AND vIRCLOGy OF HERPETICLESTCNS CF THE FACE AND ORAL MUCOUSMEMBRANES.
CRAL SURG 20:726-42, DEC 65
CAT..SCRi.'TCH DISEASE/ PATHCLOGY.*CHICKENPDX/ PATHOLOGY. *DERMATITISFERPETIFORMIS/ PATHOLOGY, *ECZEMA/
NATIONAL LIBRARY OF MEDICINE IMEDIARSI
EFFECT OF DRUGS CN HANDWRITING.
BARCLIN GS, RUPPRECHT A(ACUTE REVERSIBLE PARKINSCN1S,SYNDROME IN ENCEPHALITIS) (GER)
WIEN Z NERVENHEILK 25:76.84,, 1967ADULT, RIPERIDEN/ THERAPELTIC USE (3), DRAIN, DOPAMINE,ELECTRCENCEPHALCGRAPHY, ELECTROMYOGRAPHY. FEMALE (4)HANDWRITING, HUNAN (4), PENINGOENCEPHALITIS/ COMPLICATIONS,*PEN/NGOENEEPHALITIS/ ETICLCGY, =ARKINSONISM/ CEREBROSPINAL FLUID.PARKINSONISP/ COMPLICATIONS, *pARKINSONISM/ DRUG EFFECTS,*PHENYLACETATES/ CEREBROSCINAL FLUID
CHALLAS G, CHAPEL JL, JENKINS RLTCURETTE'S DISEASE: CONTRCL CF SYPPTCMS AND ITS CLINICAL COURSE:INT J NELROPSYCHIAT 3:SUPFL 1:9510. AUG 67ADCLESCENCE, CHILD, CH/LD, PRESCHOCL, FEMALE (4), *GILLES DE LATCURETTE'S DISEASE/ DRUG EFFECTS, *H.)LCPERIDCL, THERAPEUTIC USE,HANDWRITING. HUNAN (4), PALE (4). VOVEWENT DISORDERS? DRUGEFFECTS
CCLUCCI DtAPATC F(THERAPEUTIC ACTICN CF GANivA-AP/NCEETA-.HYDROXYLTYRIC ACM (GABCB)CN CHCRSA MINCR AND MODIFICATIONS IN GLANZIcANNIS WRITING TESTCURING THE TREATMENT) NIT)CSPED PSICHIAT 33:12231. ..AN -JUN 65ADOLESCENCE, CPILD. *CHCREA/ DRUG r.'"EqAPY. FEFALE (4),*HANDWRITING. HUNAN (4), *HYCRCXYBuTyRATES/ THERAPEUTIC USE,MALE (4)
gcwNINE RW, EEERT JN, ECRUCHM jKTEMPORAL CHANCES IN HANDWRITING, SIZE, LEVEL OF PREPORBID SOCIALFUNCTICNING AND INTELLECTLRAL LEVEL DURING TREATMENT IN ACUTESChIZCPHRENICS.J NERV PENT CIS 142152633, AN 66ADCLESCENCE, ADULT._ *CHLCRPPOPAZINE/ THERAPEUTIC USE,*FLUPHENAZTNE/ THERAPEUTIC USE. *WINcwRITING. HUMAN (4). MIDDLEAGE. ,PSYCNCLOGICAL TESTS, *SCHI2CPHNE^/A/ DRUG THERAPY.*SCHIZCPHRENIC PSYCHOLOGY, *TRANCULIZING AGENTS/ THERAPEUTIC USE
FIGURE 3.Demand Bibliographies.
e
0
tNATOMY X HISTOLOGY-ORt.G EFFECTS
1 1
SVPRYCLOGY 1 0 1
mETAeOLISN 1 1 2
PNySIOLOGY 5 0 5SURCERY 1 0 1
MESENCPYPCA 2 0 2-DIAGNOSIS 1 0 1
THERAPY 2 0 2MESENTERIC ARTERIES 2 2 . 4
DRtG EFFECTS 2 0 2cHYSIOLOGY 1 0 1
PPYSICFATHCLOGY 1 C 1
DADIOGRAPHY 1 1 2St.!'4'GERY 2 0 2
MFSEMTERTC CYSTPIACNCSIS 1 0 1
RATPOLOY 0 1 1
StIRCERY C 1 1
MESETERTC VASCULAR OCCLUSIONCO''PLICATIONS 1 0 1
RADIOGRAPHY 1 0 1
VESE*TERIC VEINSDR C EFFECTS 1 0 1
YESE'TERY 2 4 6- AWT0PY % HISTOLOGY 1 0-
CYTCLOIY 1 1 2MICROBIOLOGY 0 1 1
'AT; -OLOGY 2 2
mESOCOLONAWTOMY % HISTOLOGY 1 0 1
MEtCDERM 0 1 1-.
DRt,C EFFECTS 1 0 1
SZYMOLOGY 1 0 1
GRC4TH X DEVELCPMEMT 1 0 1
PHYSIOLOGY 1 0 1
TRANSPLANTATION 1 0 1
mESONEPHROmA 0 -
CRUC THERAPY 1 0 1
NESOTHELICmA 1 0 1
DItENCSIS 2 2 4DRUG THERAPY 1 0 1
MESTRANOL 0 1 1
AD? INISTPATION % DOSAGE 1 0 -1-ADVERSE EFFECTS 2 0 2
PHARmACODYNANICS 5 0 5
THFPAPETIC USE 1 0 1
mETARCLIC DISEASES 2 0 2
0/AGNCS/S 2 1 3-ETTCLOGY 1
FIGURE 4.MeSH Statistical Report.
cessing, and a satellite Honeywell-200 com-puter used for the slower input/output oper-ations.
MEDLARS publications are prepared forprinting by the Photon-900 Computer Photo-typesetter referred to by Library staff asGRACE (GRaphic Arts Composing Equip-ment). GRACE is an off-line photocomposerthat prints from a font of 226 dharactersonto nine-inch-wide positive photographicfilm or paper. It operates at a speed of 1.7lines or approximately 300 characters persecond. The character set includes a 6-pointfont, regular and boldface, upper- andlowercase, with a large subset of specialcharacters, including diacritical marks forforeign languages; a 10-point font, upper-case, boldface; and a 14-point font, upper-case, bold.
195
Although the MEDLARS equipment hasperformed well, continued technological ad-vances in the computer field, coupled withworkloads approaching machine capacity,have required the Library staff to begin afeasibility study for a new configuration ofequipment and related programs.
StaffAs of January 1, 1968, approximately 90
people were required to operate and manageMEDLARS*. Figure 6 shows how this staffwas organized, along with the number ofprofessional and non-professional personnelin each section.
*This does not include the staff of the CatalogingSection, Technical Services Division (15 profession-al, 10 clerical). See Chapter VIII on the CurrentCatalog.
9
MEDLARS sysTEm OVERALL DATA FLOW CHART
MEDLARSDICTIONARY
TAPE
COMPUTER
INPUT
PROGRAMS
COMPRESSED
CITATION
FILE
1.0
DEMAND
SEARCH
REQUESTS ISEARCH &RETRIEVALPROGRAMS
f
REPORT
GENERATOR
PROGRAMS
DEMAND
BIBLIOGRA-PHIES
RETRIEVAL SUBSYSTEM
INDEXED
CITATIONS
SEARCH &
RETRIEVALPROGRAMS
INPUT SUBSYSTEM
PUBLICATIONFORMAT
PROGRAMS
MEDLARS
PUBLICATIONS
PUBLICATIONSEARCH
SPECIFICATIONS
IHOLD FORANNUAL
CUMULATIONOF INDEXMEDICUS
MEDLARS
PUBLICATIONS
TAPES
FIGURE 5.
PUBLICATION SUBSYSTEM
TABLE 2.MEDLARS Data Processing Equipment
Qty Manufacturer Description
INPUT EQUIPMENT
15 Friden, Inc.3 IBM1
211
1 Diebold, Inc.
It
II,
17
77
COMPUTER EQUIPMENT
1 Honeywell, Inc.
1 Honeywell, Inc.
OUTPUT EQUIPMENT
1 Photon1 Eastman-Kodak
Flexowriters (punched-paper-tape typewriters)024 Card Key Punches026 " " Punch056 " Verifiers557 Interpreter082 Card Sorting MachineCard Tub File used for pre-punched demand search input cards
H-800 Computer with:Central Processor
Magnetic Tape Control & Tape Switch Unit%" Tape Transports (7)
8,192 48-bit Word Memory ModuleH-200 Computer with:
Central ProcessorMagnetic Tape Control Unit
3/2" Tape Transports (2)8,192 6-bit Character Memory Module
900 line-per-minute Printer and ControlPaper Tape Reader and Control Unit
Card Reader/Punch and Control UnitOn-Line Adapter for memory-to-memory connection with H-800
"900"Computer Phototypesetter (magnetic-tape-driven)Versamat Automatic Film Processor
The Bibliographic Services Division in-cludes trained literature analysts who indexnew journal articles and formulate demandsearches for the computer, as well as theMedical Subject Headings staff responsiblefor maintaining and updating the mastervocabulary of terms on which the entire sys-tem is based. The Office of Computer andEngineering Services* is responsible for com-puter programming and program mainte-
*Called Data Processing Division to September1965; Information Systems Division to January1968.
nance, operation of data processing equip-ment, and systems analysis and design.
The personnel described in Figure 6 areonly those directly involved in operation ofMEDLARS. There are many other groupsin the Library which perform important re-lated activities, such as selection and acqui-sition of books and journals, administrativeservices, and research and development.
Those readers who are not interested in adetailed description of how MEDLARS op-erates may omit the next four chapters andturn directly to Chapter VII, OperatingExperience: 1963-1967.
11
Medical SubjectHeadings Section
6Professionals2Nonprofessionals
DIRECTOR
Associate
Director,Library
Operations
BibliographicServices
Division
Index Section
22Professionals2 Clerical
Search Section
11 Professionals4Clerical
12
Office ofComputer andEngineering
Services
Computer Applications Sect.
8 Professionals(Programmers)
Systems Analysis Sect.
5Professionals1Clerical
Computer Operations Sect.
1Professional24Technician/Clerical
FIGURE 6.NLM Staff Working on MEDLARS.
Chapter III
INPUT SUBSYSTEM
The Input Subsystem is the functional por-tion of MEDLARS concerned with selectionof journal articles, indexing, conversion tomachine-readable form, and input to thecomputer for storage on magnetic tape.
Coverage and Selection of JournalsThe National Library of Medicine currently
receives between 18,000 and 19,000 differentserial publications of all types, ranging fromserious scientific journals to newsletters andpopular works. The contents of approxi-mately 2,300 biomedical journals are indexedfor input into MEDLARS from the totalnumber of serial publications received.
Selecting journals to be indexed is verydifficult. The Library is aided in this taskby the Committee on Selection of Literaturefor MEDLARSan external advisory groupof experts representing various medical spe-cialties, medical librarians, and medical edi-tors. The criteria for selecting journals forMEDLARS include: (1) coverage of a crosssection of the various disciplines in the healthsciences; (2) selection of the higher-qualityjournals ; and (3) coverage of a cross sec-tion of foreign publications.
Journals selected are divided into twogroups based upon the scientific significanceof the material published ; a depth-indexinggroup (journals that regularly carry reportsof greater significance) and a non-depthgroup (journals containing material of lessersignificance). The depth journals are indexedin much more detail than the non-depth, asdiscussed below. From the total collection ofjournals selected for MEDLARS, approxi-mately 55% of the articles are in English;45% are written in over thirty foreign lan-guages.
Figures 7 and 7a illustrate the flow of
materials through the MEDLARS InputSubsystem.
Journals received by the Library are for-warded to the Selection/Acquisition Sec-
tion of the Technical Services Division. TheMEDLARS journals are matched against aseparate MEDLARS Serial Record. Afterserial checking, the cover is rubber-stampedand the MEDLARS Journal Title Code iscopied from the Serial Record onto the coverof the journal, along with the volume andissue number (see Figure 8). The MEDLARSjournals are batched and forwarded to theIndex Section, Bibliographic Services Divi-sion, several times each day.
Medical Subject HeadingsBefore continuing with a description of the
information flow through the Input Subsys-tem, it is important to describe the controlledvocabulary on which the entire systeni isbasedthe master file of Medical SubjectHeadings (MeSH). This section of ChapterIII will describe the structure of MeSH.Procedures for maintaining and updating themaster MeSH file are covered in Chapter VI.
MeSH is a controlled, yet dynamic, list ofapproximately 7,200 subject headings, eachdefined to represent a specific concept in thebiomedical field. In addition to the mainheadings described above, the MeSH file alsocontains some 250 geographical headings(place names); some 500 provisional head-ingsnew terms being considered for inclu-sion as main headings ; and 60 subheadingsused in connection with main headings atthe indexer's option. In the example KID-NEY NEOPLASMSSurgery, the first term,KIDNEY NEOPLASMS, is the main sub-ject heading; "Surgery" is a subheading.
13
MEDLARS INPUT SUBSYSTEM FLOW CHART
JOURNAL
ISSUES
RECEIVED AT
NLM
MEDLARSJOURNAL?
YESCHECK AT
MEDLARS
SERIAL
RECORD
APPLYMEDLARS
STAMP TO
COVER
ASSIGN
JOURNAL
TITLE CODE
(JTC)
NO
MEDLARS
NON-
PROCESSING
TECHNICAL SERVICES DIVISION
DESCRIP-
TIVEINDEXING
SUBJECTINDEXING
1
TRANSLATETITLETITLE IFNECESSARY
CHECK &
REVISE
INDEXING
I
DISTRIBUTETO
INDEXERS
FINALCLERICALWORK
TRANS-
LITERATEIFNECESSARY
BATCH &FORWARD
FOR
PUNCHING
BIBLIOGRAPHICSERVICES DIVISION
I
JOURNALS
INDEXER
DATA
FORMS
PUNCHED
PAPER TAPE CORRECTIONS
INPUTPUNCHING/
TYPING
14
HARD
COPY
CORRECTIONS
SIGHT
VERIFYHARD
COPY
SPLICE& BATCH
TAPESORIGINALTAPES
OFFICE OF COMPUTER & ENGINEERING SERVICES
FIGURE 7.
MEDLARS INPUT SUBSYSTEM FLOW CHART
WWITAPES
CORRECTIONMATCHING
& INPUTEDITING
1.4\RETURN TO
INDEX SECTION
OR INPUT
FOR CORRECTIONS
111112tliir
INPUT &ERROR
LIST
CHANGES TO
JOURNAL RECORD FILE (JRF)
VALIDATEJOURNAL
TITLECODES
UPDATEDJRF
INTER-MEDIATE
INPUTTAPE
INPUT MODULE
4-MEDICAI
SUBJECT
HEADINGS
FILE
VALIDATESUBJECT
HEADINGSUPDATED
MeSH
FILE
INTER-
MEDIATE
INPUTTAPE
COMPRESSED
CITATION
FILE
INI100.-FORM
UNIT
RECORDS
/
UPDATED
COMPRESSED
CITATIONFILE (CCF)
*These files are found on the Medlars Dictionary Tape.
FIGURC la.
15
CELLBIOLOGY
FIGURE 8.--Stamped MEDLARS Journal.
MeSH is published once a year in an al-phabetic and categorized arrangement (seeFigure 9). In addition to this published ver-sion, several computer listings are printedeach year for use internally by indexers andsearch specialists at NLM.
Keeping MeSH current with new conceptsand terminology in medicine is a demand-ing task. The professional staff memberscharged with keeping MeSH up-to-date havereceived assistance from outside experts invarious medical specialties. Although outsideassistance has been helpful, a recently corn-pleted MEDLARS evaluation (described inChapter VII) has shown that more relianceon actual term usage in the literature and in,search requests, with less reliance on com-mittees of experts, will help to further im-prove the vocabulary.
In addition to the alphabetic terms, MeSHhas also been structured into a hierarchicalclassification, or "tree structure," to facilitate
16
search and retrieval. A sample page fromthe tree structure is shown in Figure 10.
MeSH is not a new vocabulary; rather, ithas been developed over the years with theprintea Index Mcdicus and its predecessorpublications. Mf content of the vocabularyrelates to usage of terms in the literatureitself. This necessarily means that the listof terms must be dynamic and change tomeet new concepts in the field of medicine.The dynamic nature of MeSH during theMEDLARS years is dramatically shown inTable 3.
Table
Changes in the Medical Subject Headings File
Item 1964 1965 1966 1967
Main Headings 5812 6345 6566 6800
Geographical Headings 149 144 175 264
Provisional Headings 300 612 531 434
Subheadings 0 0 42 53
Indexing and RevisionJournals received from the Selection/Ac-
quisition Section (see flow chart, Figure7) are given first to a highly trained clericin the Index Section, who verifies the JournalTitle Code and transliterates the title andnames of authors for all journals printed inCyrillic alphabets. This clerk also separatesthe journals into categories : those to be in-dexed in depth, the non-depth journals, thoseto be handled on a "rush" basis for process-ing, and those to be selectively indexed formedically related papers only. The journalissues then are distributed to the profes-sional indexers, taking into consideration thespecial subject or foreign-language skills ofeach individual.
The indexers prepare an Indexer DataForm (see Figure 11) for each article in thejournal?. Most journals are indexed cover-to-cover for all substantive articles. These mayOD occasion include, in addition to originalarticles, letters, editorials, biographies, andobituaries. A few journals, such as Scienceand Nature, are indexed selectively only forarticles in the field of medicine.
The indexer first scans and evaluates thearticle to find out what it is about and what
O ng Effects (kik P. G1 Used ter the effect oddrugs and chemicals on an organ, region. ortissue; me bacteria. vine's. Bast. EA Santa.tetua. tethers. ad phyolological processes,
Doug Therapy (C,F) Used ter the treatment orthe Proecbitc c dilbea by the strainfatrationof 1rusa, ckemicals and other therapeuticoyes, suck ss smiUotics, biolegicale. andBoot 'Risotto. EL:hides diet therapy.
Eb.Cation Used for education andinlning programs la specialty fields.
Embryology (A. el Utel ler embryolsgic andIota) develomenL
Pothytnology II, B. C. F1 Used for enzyme owl.les, Including Inhibittec, kinetics. etc., inHaim. piano, microorganisms, and Inverse-bratea; else used for studies of enzymes dur-Ing course of disease. Does not Includediagnostic easy= teats.
Cabin), (C.F) Used for the Causative factorsof disease including microorganisms or otherdisease factors. Includes (actors associatedwith an increased flak a disease such asenvironmental. social, and personal (actors;Includes patitogereath.
Gsvtic :C. Fl - Used (Sr studio egenetic both deaf sae: studies Al diseaseoccurring le (aridly zroupe 'Jane to infection.common family dietary habits or en.lrwunentaltatters: also thdicstes hereditary disease.
Growth & Development (A. B) - Used for growthand development of microorganisms Indolent,,end for the postnatal development of animate.
fl::tory (C through N1- Used for the historical'INV. of any subject: Includes brief historicalnote,. Escludes case histories.
Immunology (A.B. C1 - Used for immunologicstudies of times. organs. bacteria. virmeoand host and their constituents: includes tm.munologic aspects of specific di but notImmunologic procedures used for diagnostic.preventive, or therapeutic purposes.
Inj tilts (Ai Used (Cr 'osmotic wounds awlI spates of ports of the body.
Innovation (A) Used to nerve supply of anorgan, region, or tissue.
Inrrinnenta rem (E. F. - Used for ems.ram, instruments. and equipment. their de-velopment ormoifification.thruseindtagnostic.analytical or therapeutic procedures.
Isolation 11 Purification Used tor mettodaewpiryed n *bola s pure culture or strainedmicroorganisms. basks Culture teclaits.
Hoopoes (E.P.G.H.I.L.N1 Used Or the de-mand, st7:ply, distribution. recruit:nen anduse Al personnel la the various health die.
Metabolism (6.13, C. 0, - Used (Sr metabolicprocures occurring la organs or regions, orthe *rosier* es a whole; includes the studyof metabolism In disease states; Includesabsorption, distribution. blotransformatten andetcregen of drugs; sad the absorption, Cs-Dilution, breakdown products and escretienof noulento or the I r constituents. Forbiosyntheath, enzymology and secretion. thespecific subheading it use&
Microbiology IA. C1 - tied Isr microbiologicalstudio of organs, regime and diseases.
Mortality (C.'E.F) - Used for statistics of mor-tality 121221d2,2 from disease or is s result ofprocedures
List of Subheadings
Categorized List
G - BIOLOGICAL SCIENCES
GI - General Biological and Preclinical Sciences
ANATOMYBIOCHEMISTRY (H)BIOLOGYBIOPHYSICS (H)BOTANYCYTOLOGYECOLOGYGCNETICSIMMUNITYMARINE BIOLOGYMICROBIOLOGYPHARMACOLOGYPHYSIOLOGYZOOLOGY
ACCLIMATIZATIONACCOMMODATION, OCULARACID -BASE EQUILIBRIUM(H)
Differs (H)ACTION POTENTIALSADAPTATION, BIOLOGICAL
Adaptation, PhysiologicalADAPTATION, OCULARADAPTATION, PHYSIOLOG-ICALAcclimatizationHibernation
ADOLESCENCEPuberty
ADULT (M)AGED (M)AGGLUTINATION
HemagglutinationAGING
AdolescenceAdult (M)Aged (Id)Child (M)Infant (M)LongevityMiddle t.ge
AIR MICROBIOLOGY (G3)ALLELESAMELOGENESISAMINO ACID SEQUENCEANATOMY
Anatomy, Artistic (K)Anatomy, Conurirative
376
ANATOMY (Continued)Anatomy, RegionalAnatomy, VeterinaryCytologyEmbryolHistology
ANATOMY,ANATOMY,ANATOMY,ANATOMY,ANEUPLOI
TrisomyANIMAL hANTIBODYANTIGEN.
REACTIOAgglutinaHemolysiPassiveAnaphyl
AROUSAL (BACTERIO
BacteriolGerm -FtNitrogen
BACTERIOBASAL MBINDING SIBIOCHEM
Acid-HasAminoBindingBody CoCalcifies
42
4111
BIOCHEMISTRY (H)(Continued)Enzyme Induction--,v w 1
Alphabetic List
COMPREHL0SITE HEALTH CARP.(Ni)evONSWIrrre., Fanny ca ,,,,, ,zu sne 11122[12I IMnzY /12.2T CA II 11115
CIIMPBMIMM0412222CL 551* 2212222121,
0001PEL.0117 RENA% IOR (F)row.
MLLTtMYwwosu.
IV 01421212I11.2222114212141LOILIL 0.1121120222V2Ina IT,
COMPLTERS (L)22. CONIUTI22.21=2 luno MICIAVU1112. COWL". , It
COMITTLILC. ANALOG (L)
0331 MILS. DIZITAL (L)1221... . C010.17.11.1.
MOURIATIM. rattan,,,
CONCEIT FORMITION (F)
CeNtimilme nuntlnAllem.C.e
C11.11. IVI ne/22122 ,1
037101311NTS (ite in tob.<I.. OMMALI( MitGintri 1110.
20 CM. VOW Ilt20 antra Oa
CONDITIONING, OPERANT
CONDITIONING (PNlafoLoGI) fn
12...22MITT ItiazICCON2vanrITTICI. 21
CONDIVTOMETts (HiCO22.142211 23
to2nusoo ... CATIMIK1...alt111222LIKI I
CONFLICT (Psi alowcY) (F)
CON(X1 RID (Dil)
CONGRLCSFA (Na)IV (WM DK= III.II VITI22211002L coorzureg .1.
CONSIZATION. GENETIC (GI)II MOM/. /VIM .411
CON2L'NCTIVA (Al)
CON(LTSCTIT(TIS (Cl))
coNiuncnvrns. INCLUSION(Cl. ClIl2 SLIVol.alx..32222.1 CI"
CONNECT:YE TISSII (A10)
dim
CONNECTDE TIS01 E CELlN (All)
22221 M.la myrta12027n021S2
C2121, w mgC
CONNANG1INITT (GIIan .1.000 CUM .C11.as 'Ma .2.1tan 17111.110.C.C11
CONSOMT orft a, in
CONSC1OFFNESC (F)ainMNw.Ilnl2a.pin
CONCTIPATION (C41
feNSTITITIO. 10111 0.1.1IOV
122.T11(11IL halt Ocr, 6.4IlTkrIZIMIT
rov.ssesswsv00051
CON0I NIER DRUtNIZATIO 44 (021
CONCF0IER NATICFACTION (0))
CON0I 00117014Y RE1141 101 (F)
CONTACT LEN.ZE0 )E2)
7 Fa At= or. StMt
(.71Palf, AZ/ ...ow..(1,01.11rItInt to IA.TM 22222
CO %TR trEPTIOV 4E2)
CONTRACEPT1lE ACENT74 111111
CONTRACEPTI% E DE% KIN (Eg)
CPR. ftTRII 1.1111WRIM11TIA Vita I.t OtIfICF.S
CONTBACEITITE0. ORAL II/S.1)121
CONTRACTFRE (C3)
CONTRACT MEDIA mu)no DI r.01112.T. 0111IV 2.02.1UOIK2IVonII I/MOGI. 221f II
CONTINION1 (Cm
CONVALLSCI.NCE (C17. NI)
CONVALLIRIA (Bt. DI)
tIont122.2. tog
FIGURE 9. Specimen. MeSH Pages.
17
18
N2 - HEALTH FACILITIES, MANPOWER AND SERVICES
FACILITIES MANPOWER SERVICES (NON MESH)HEALTH FACILITIES
AMBULANCESBLOOD BANKSDENTAL CLINICS
HOSPITAL DENTAL SERVICEEXTENDED CARE FACILITIESHEALTH FACILITY SIZEHOSPITAL DEPARTMENTS
HOSPITAL CENTRAL SUPPLY'HOSPITAL COMMUNICATION SYSTEMSHOSPITAL DENTAL SERVICEHOSPITAL EMERGENCY SERVICEHOSPITAL FOOD SERVICEHOSPITAL HOUSEKEEPINGHOSPITAL MAINTENANCEHOSPITAL MEDICATION SYSTEMSHOSPITAL NURSING SERVICEHOSPITAL OUTPATIENT CLINICSHOSPITAL PHARMACY SERVICEHOSPITAL PSYCHIATRIC DEPARTMENTSHOSPITAL PURCHASINGINTENSIVE CARE UNITSLIBRARIES, HOSPITALOPERATING ROOMS
HOSPITAL PLANNING AND CONSTRUCTIONHOSPITALS
HOSPITAL DEPARTMENTSHOSPITALS, CONVALESCENTHOSPITALS, GENERALHOSPITALS, PSYCHIATRICHOSPITALS, SPECIALHOSPITALS, TEACHINGHOSPITALS, VETERANS
LABORATORIESLEPER COLONIESNURSING HOMESPHARMACIESPOISON CONTROL CENTERSR2HABILITATION CENTERSRESIDENTIAL FACILITIES
HALFWAY HOUSESHOMES FOR THE AGED
SHELTERED WORKSHOPSTISSUE BANKS
HEALTH MANPOWERDENTISTS, WOMEN
ADMINISTRATIVE PERSONNEL
N2.N2.18.N2.18.3.N2.18.6.N2.18.8.N2.18.8.1N2.18.9.N2.18.12.N2.18.15.N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.15.1N2.18.21.N2.18.24.N2.18.24.1N2.18.24.1N2.18.24.1N2.18.24.1N2.18.24.1N2.18.24.1N2.18.24.1N2.18.27.N2.18.30.N2.18.33.N2.18.36.N2.18.39.N2.18.45.N2.18.48.N2.18.48.1N2.18.48.1N2.18.51.N2.18.54.N2.36.N2.36.1.1N2.36.3.
N2.18.15.1
N2.18.24.1
N2.18.8.1
N2.72.23.1
N2.72.18.1
N2.72.51.1
L.72.37.1
N3.39.45.
N2.18.15.
M.25.32.
M.1.
M.109.21.
INDICATES PROVISIONAL HEADING
N2-1
FIGURE D.Specimen Page, MeSH Tree Structure Listing.
Aononymous0non-std dotePnon-std pagination
PAGINATION OPT. PAGINATION
AUTHOR DATA (Print)
AUTHOR DATA (Sort)
TITLE (English or English Translation)
TITLE (Vernacular or"Transliterated Vernacular)
IM NIM MAIN HEADING 'subheading IM NIM CHECK TAGS
REVIEW--- References INFANT, NEWBORN (to 1 mo.)
PREGNANCY INFANT (1-23 mos.)CHILD, PRESCHOOL (2-5 yrs.)CHILD' (6-12 yrs.)ADOLESCENCE (13-18 yrs.)
ADULT (19-44 yrs.)MIDDLE AGE (45.64 yrs.)AGED (65- yrs.)
CATS
CATTLEDOGS
GUINEA PIGS
MICE
RABBITS
RATS
HISTORY OF MEDICINE
ANCIENT
MEDIEVALMODERN
15TH CENT.liTH CENT.
Mg RACIAL AND ETHNIC GROUPS 17TH CENT.18TH CENT.
1 CENT.19THTT CC NN
GEOGRAPHIC HEADINGSEX,:,: HISTORICAL ARTICLE
3:g.: HISTORICALBIOGRAPHY: '',i't CURRENT BIOG-OBIT
,::::t r'''''s r ANIMAL EXPERIMENTS
r/..:... ,I HUMAN
MI PROVISIONAL HEADINGS .1 IN VITROFEMALE::,:1 MALE
: S: .'5 I CASE REPORT
SEM CLINICAL RESEARCH
112aim COMPARATIVE STUDY
PHS-3266REV. 10.65
DATA FORM-INDEX MEDICUS
FIGURE 11.Indexer Data Form
19
are the most important points to be cov-ered. Subject headings and subheadings areassigned from MeSH. The Indexer DataForm includes several check tags which serveas reminders to the indexer of conceptswhich always are to be covered (e.g., agegroups, clinical report, etc.). In handling adepth journal, the indexer may use as manysubject headings as are needed to describefully the content of the articles. When in-dexing a non-depth journal, the indexer islimited to subject headings that describe theprimary concepts only. As of January 1968,depth journal articles were assigned an aver-age of about 10 subject headings and non-depth journals were assigned an average ofabout 4. The indexer also assigns subhead-ings and must insure that he uses a validmain heading/subheading combination ineach case that a subheading is used.
In addition to assigning MeSH terms, theindexer decides whether each term is to be"print" or "non-print" ; that is, to be printedin Index Medicus or to be used only in theretrieval process.
Descriptive indexing by the professionalindexer includes establishment of pagination,establishment of author name format (ifnecessary), and translation of titles of allforeign-language articles.
After indexing, the journals with dataforms attached are sent to the revisers (se-nior professionals who check and revise thework of the indexers). All new and traineeindexers receive 100% revision ; the moreexperienced personnel are revised on a spotcheck basis.
After completion of work by the profes-sional indexers and revisers, the journals goto a final clerical work station, where "sortauthors" are established. Sort authors arerequired in cases where the computer is notable to follow its normal collating sequence inpreparing alphabetic author lists (e.g., St.Lawrence to sort as Saint Lawrence).
In addition to performing regular produc-tion indexing functions, the Index Sectioncarries a heavy training load, both for itsown new staff members and for personnelworking at Decentralized MEDLARS Sta-tions and others outside NLM.
The original MEDLARS philosophy was
20
to perform all indexing centrally with NLMstaff. However, the massive volume of workto be done, coupled with rapidly increasingbacklogs, caused Library management to re-consider this policy and begin to use outsidecontracts for some of the indexing work.Decentralized indexing is now underway atsuch places as Keio University in Japan;the MEDLARS Stations at Harvard, the Uni-versity of Alabama, and the University ofColorado; and in Israel, using PL 480counterpart funds. Private contractors havealso been used. This decentralized indexinghas proven quite effective.
In order to insure standardization of index-ing and to facilitate the training describedabove, a MEDLARS Ind1xing Manual wasprepared. All decentralized indexing is sub-ject to revision.
After completion of all Index Section tasks,batches of journals and data sheets are for-warded to the Office of Computer and Engi-neering Services for data punching and com-puter processing.
Input Preparation and VerificationPunched paper tape was selected as the
input medium for indexed citations enter-ing the computer system because of its ver-satility in handling variable-length recordsand its ability to input 88 different charac-ters (an important consideration for MED-LARS publications). Punched cards andoptical scanning of citations prepared by atypist (not scanning of the original journalarticle) were also considered, but, after acareful analysis, paper tape seemed to pro-vide the best approach.
As seen from the flow chart, Figure 7, theclerk-typists in the Input Unit receive jour-nals with Indexer Data Forms attached forinitial processing. For each journal received,the typist first punches a Journal Identifica-tion Record which contains the Journal TitleCode, publication date, volume and issue num-ber, Indexer Identification Code, and TypistIdentification Code. The typist then punchesone tape record for each indexed article inthe journal. This record includes the title ofthe article in the vernacular or transliter-ated vernacular (except for oriental articles,which include only a title translation) ; the
00
00.00
00000000Soo
000
000
40 00 00 011! 4D .00 ', 00 , 0 0 ,
I 000000 (to, 00000000000 oloweeam* 0000 44004,041140410*'' 0 Iv
oolv ofto 00 00, -400000 0.4 1 A1 :00. ,,o'
. /
7Q6 4312 225-2
*11 -22 *Korting aws Ntirnberger F, +NURNBERGER F: [Necrosis lipoidica] (Ger)Necrosis lipoidica
Paper Tape Strip
*13 Jun 66
T/NECROBIOSIS LIPOIDICA+PATHOLOGYSKIN+PATHOLOGYNECROSIS+PATHOLOGYLEGMIDDLE AGEHUMANFEMALECASE REPORT
R./
C/ *123-35 *A/ Simon N, Berencei G: [Contributions on the biological properties of Mycobacteriumtuberculosis on the basis of the sensitivity of lupus strains] (Ger)11/ Beitrige zu den biologischen Eigenschaften auf Grund der Empfindlichkeit der Lupus-Stimme
TiLUPUS+MICROBI0L0GYTUBERCULOSIS, CUTANEOUS4MICROBIOLOGYMYCOBACTERIUM TUBERCULOSIS+DRUG EFFECTSSTREPTOMYCIN+PHARMACO FlexowriterISONIAZID+PHARMACOISONIAZID+THERAPEUTIC USESTREPTOMYCIN+THERAPEUTIC USELUPUS+DRUG THERAPY Hard CopyTUBERCULOSIS, CUTANEOUS+DRUG THERAPYDRUG RESISTANCE, MICROBIA
FIGURE 12.Paper Tape Strip and Flexowriter Hard Copy.
translated title of foreign works; names ofall authors or editors; the abbreviation forthe language in which the article is writtenif it is a foreign work; the name of thesubject(s) of a biography; the paginationof the .article within the journal; and thedescriptive tags assigned by the indexer, in-cluding main subject headings, subheadings,geographical tags, and provisional headings.Those subject headings to "print" in IndexMedic= (headings under which the articlewill be cited in the published index) arepunched in red-shift code and non-IndexMedicos headings are punched in black-shiftcode.
The paper tape typewriters (Friden Flexo-
writers) operate under the control of a pro-gram tape which assists the operator in
operly sequencing and identifying the datafields for the computer. The Flexowriterssimultaneously produce a proof copy on con-tinuous paper, along with the punched tape(see Figure 12).
The typist normally treats one journalissue as a batch of work for further process-ing, except for those journals with a veryfew articles, in which case several issues maybe combined into one batch. The entire pack-age (journal issue [s], Indexer Data Forms,Paper tape, and Flexowriter hard copy) isthen forwarded to the proofreaders.
21
The proofreaders sight-verify all descrip-tive elements of each citation on the hardcopy (the subject headings are validatedlater during computer processing). Errorsdetected by the proofreaders are noted, withappropriate corrections on the hard copy,which is then returned for correction typing.The proofreaders detect one or more errorsin approximately 20% of the articlespunched.
Paper tape creates some problems in mak-ing corrections. Rather than attempt to cor-rect the original tape itself by duplicationor splicing, correction record (s) are punchedwhich identify the journal article, specifythe data field in error, and give the correc-tion data. The correction record (s) are sub-sequently matched and applied to the originalrecords during computer input processing.
All original input tapes and correctiontapes are spliced together at the end of theday and batched for the daily input computerrun. Note that correction tapes must be en-tered into the same computer i'un as theoriginal data tapes to which they apply.
As of January 1, 1968, the average num-ber of articles indexed and entered into thecomputer per day was approximately 700,or about 15,000 new items each month. It isobvious from these figures that MEDLARSmust operate as an organized production sys-tem in order to maintain level workloads.
Input ModuleThose computer programs which input the
new journal articles and update the mastermagnetic tape file of citations are referredto as the Input Module. The Input Moduleconsists of four major segments or sub-modules (see flow chart, Figure 7).
The first submodule reads in the papertape (original and correction records),checks to see that only valid paper tapecodes are present, and copies the recordsonto magnetic tape. The magnetic tape rec-ords are then sorted so that corrections arematched to the original typed citation. Thecorrection data is then applied to produce a"clean" record on magnetic tape. Any errorsdetected by the computer (invalid codes orcorrection records which do not match) are
22
FIGURE 13.Paper Tape Being Loadedonto Tape Reader.
printed on an error listing and these recordsare rejected.
The second submodule validates the Jour-nal Identification Record for each journalissue in the batch. The Journal Title Codesare matched to a master Journal RecordFile (JRF) on magnetic tape. Note that ifthis code is in error, all articles from thatissue must be rejected and printed on theerror list. The subsequent correction, retyp-ing, and re-entry of the material into thecomputer is a costly process. The JRF pro-vides the Journal Title Abbreviation whichis added to the citation record at this point.
This submodule also is used for updatingof the Journal Record File. Changes to thefile (additions, changes, deletions) are en-tered via punched cards and the file is up-dated during input processing. The JRF alsoincludes the full title and place of publica-tion, which are used in printing the annualList of Journals Indexed in Index Medicus.
The third major submodule provides vali-dation of all subject tags in the citations.The computer splits each citation into several
Honeywell FORMAT MAP B.3
elatz'aapar, Layout Form Item Design
TitleCompressed Citation Data Record
Prepared by For Program
By ProgrammerChecked by
DateRemarks.*
indicates a variable number of entries
numbers in parentheses following field names indicate number of bits for that field.
ModificationPage of
12 11 10 9 8 7 0 4 3 2
Place of Publication (17) Year of Publication(12)
Open
3)
Item Word ^ount (16)
Citation Number (20) Open
(4)
Journal Title Code (18) LanguageCode (6)
Form (7) Subform(5)
IndexerID No.(6)
TypistID No.(6)
11 i
Date of Entry (13)
I.-
en(2)
TA N(2)D
No. ofTag Wds.(6
Main Heading Code (17) SubheadinCode (6)
GagsMeSH Classification No. (21)
Author(s)
778
Title 378 Journal Title Abbreviation 568 Volume No. 578 Pagination
768
Publication Date 368 Vernacular Title 538
1
No. of References
1.., i 4
Issue No. (12)
528
,
Open)
t i 1 ' 1
1.1rr--End of Item (32)
ORTHO 1
ORTHO 2
1.- ; I--4
--r--11 I 1
1 ' 1 1 4-1'7END OF RECORD
1' 1 '
--7T rI
i
i ) i
I
1 '1-1-11I'l. I I
I
i
I - i ' I I
i
I ' 1 ' 1 I I
.1.1,1.1,1,tit.I.I.I.L.1 1 I ' I r'
1021.3578 7 6 5 4 3
FIGURE M.-CC? Magnetic Tape Record Layout.
2
23
small records containing: one alphabetic sub-ject heading and the related subheading (ifany), and a Citation Identification Code.These records then are sorted into alpha-betic sequence by subject heading andmatched against the Master Subject Head-ings File on magnetic tape. Those headingswhich do not match the computer masterfile are printed on the error list and rejected;however, the citation is not rejected unlessall Index Medicus print headings attachedto it are rejected. Main heading/subheadingcombinations also are checked for validity.
Each time a match occurs, the masterMeSH list on tape is updated by adding oneto a tally of the frequency of term usage.This tally becomes important later in theRetrieval Subsystem. The alphabetic subjectheadings are replaced by code numbers pro-vided by the master file. The coded subjectrecords then are sorted back into citationnumber order, matched to the original cita-tions, and complete citation records are re-built.
The fourth and final submodule builds thecompleted MEDLARS citation record onmagnetic tape. The record is referred to asthe MEDLARS Unit Record and the tapefile itself is called the Compressed CitationFile (CCF). A record layout of the data
24
elements on the CCF is shown in Figure 14.The CCF is a highly compact tape file
which serves as the major store of informa-tion. in MEDLARS, for both retrieval ofindividual demand bibliographies and prep-aration of published indexes. The file is se-rial, with records in sequence by computerentry date. On January 1, 1968, the CCFcontained 645,751 citations on 25 reels ofmagnetic tape.
All errors detected by the computer editingroutines are printed on an error listing whichis returned to the Head of the Index Sectionwho initiates correction action. (Note thatan error could have been made by either theindexer or input typist.) Approximately 10%o the records entering the Input Modulecontain errors which are detected by thecomputer editing routines.
The various checks built into the InputSubsystems (revision of indexing, proofread-ing, computer edits) provide a good measureof control over the accuracy of data in theMEDLARS file. Nonetheless, some errors doget through, and these are usually picked uplater when they appear in a publication ora demand search.
The Input Subsystem provides the rawmaterial used for retrieval and publication,described in the next two chapters.
Chapter IV
RETRIEVAL SUBSYSTEM
The Retrieval Subsystem of MEDLARSis concerned with matching requests for in-dividual bibliographies to the citations onthe Compressed Citation File. This subsys-tem combines the search formulation talentsof trained search specialists with the rapidmatching and retrieval capabilities of thecomputer to produce demand bibliographies.
Request Analysis and SearchFormulation
The demand search cycle begins with re-ceipt of requests for MEDLARS searchesfrom medical educators, practitioners, andresearchers. The searches are requested fora variety of purposes (e.g., to determine thestate of research in a particular field, to as-sist in the preparation of review articles, tohelp solve a clinical problem). Most searchrequests processed at NLM and the 'U.S.
MEDLARS Stations (at UCLA, Harvard,Ohio State University, and the universitiesof Colorado, Alabama, and Michigan) arereceived from research workers and edu-cators, with about 95% originating in theUnited States.
A clear, concise statement of the searchrequest is important for effective retrievalin MEDLARS (as in any other reference orretrieval system). Requests are made onMEDLARS Search Request Forms (see Fig-ure 15) and may be submitted through localmedical libraries or directly to NLM. Inorder to improve search requests and opti-mize use of the service, a booklet has beenprepared to assist users in understanding thecapabilities, limitations, and conditions un-der which searches are accomplished.*
In submitting his search request, the useris asked to present a detailed, specific state
meat of requirements, to cite pertinent cita-tions that he may know of, to estimate thenumber of citations he would expect to re-ceive, and to state any restrictions to be.placed on the search. He also indicates whichlanguages he wishes included in the searchand whether he wants his final bibliographyprinted on 81/2" x 11" paper or on 3" x 5"cards.
The search specialist (or "searcher") is ahighly trained intermediary who must inter-act with both the requester and the computer.This requester-searcher interaction can oc-cur at several different levels, including: (1)request form or letter received in the mail,with no direct contact; (2) request formreceived in the mail after requester has haddiscussion and advice from his local librar-ian; (3) request received by telephone call;(4) request received during direct inter-view with the requester. Searchers are en-couraged to clarify requests and make lib-eral use of the telephone before beginning toformulate the search.
After the searcher has analyzed and clari-fied the request, he begins building his searchformulation, which is used as the basis forcomputer matching and retrieval. Figure 16is a sample search formulation. The formu-lation consists of two major parts: a list ofthe search elements to be used, and fromone to three search equations showing thelogical relationship between elements' in thesearch.
The main elements used in searching are
*A Guide to MEDLARS Services, U.S. Dept. ofHealth, Education, and Welfare, Public Health Serv-ice, National Library of Medicine, Bethesda, Mary-land, 1967.
25
070852
MEDLARS SEARCH REQUEST
U. S. DEPARTMENT OFHEALTH, EDUCATION, AND WELFARE
PUBLIC HEALTH SERVICENATIONAL LIBRARY OF MEDICINE
MAR ': 1968DATE6 March 1968
1. INDIVIDUAL WHO WILL ACTUALLY USE THE BIBLIOGRAPHYCaptain William Mann
TITLEDermatologist
ORGANIZATIONWomack Army Hospital
ADDRESS
Fort Brag' N.C. 28307
2. REQUEST SUBMITTED BY (If different from above):Mrs. C. Edwards Medical Librarian
3. DETAILED STATEMENT OF REQUIREMENTS (Please be as specific as possible as to purpose, scope, definitions, limitations, etc.)
Dr. Mann is interested in all of the articles on Molluscum Contagiosum
particularly those concerning innoculation studies and clinical studies of the
naturally acquired infection in man.
Any references under venereal disease and Molluscum acquired as a. venereal
disease would be pertinent.
4. TITLE OF PROJECT FOR WHICH SEARCH IS REQUESTED (Omit if not applicable):
4.6.'4\
''''*".
",'se. iggs..e.0'?, 4-
S. MEDICAL TERMS PERTINENT TO REQUEST (Optional). DESCRIPTIONS CURRENTLY USED IN MEDLARS ARE PUBLISHED
IN Medical Subject Headings, Part 2 of the JANUARY Issue of INDEX MEDICUS.
MOLLOSCUM CONTAGIOSUM
6. LIMIT
LANGUAGESTO
og ACCEPT ALL 7. PRINT SPECIFICATIONS:12 3"X 5" CARDS0 PAPER
ID ENGLISHJ FOREIGN (Specify):
PHS - 4667-1 (REV 5.66) HEW-Lex
FIGURE 15.Specimen MEDLARS Search Request.
26
REQUEST NO.
070852DEMAND SEARCH FORMULATION RECORD
TITLE
Molluscum Contagiosum.
.41 Awl
MAR 1 719680A1EMarch 15, 1963
.1111MINI.4111111
1144 17 18 19..31 i
ELEMENTSjJ<I-
1144 17 18 1941
ELEMENTS
,-
<I'-
*0XMt21.Wy
..1
5W
JW
J
A>5-m<o(1°2
0riTJ5-.Win
;
ii:
xW
W>wJ
Mup-W"T<05022.'
M1M2
M5M6M7M8M9
S1
S2
S3
S4
E
Moluscum ContagiosumMolluscum Contagiosum Virus
Clinical ResearchEpidemiologyImmunityImmunologyVenereal Diseases
etiologyimmunologyoccurrencepathogenicity
rI-oWN
EL.M. ELEMENTS A, 4 1, N, V. X, AND SUMMATIONSSYRODL
7
3-
11.14 33.36 37.40....
41.44 4540 49.563.
114 4 33 -36 37-40 41.44 45.48 49 -56
M25 SUMM30 SUMS5 SUM
M1
M5SI
M2
M9S4
UWo
ELEM. REQUEST STATEMENTSSYMBOL
7 9.10 11.80 COLUMNS FOUND
al.4 01 T/3.
5 01 M30 + s5.
....i.---M .
DS MODULE COMMENTS ROMODULE COMMENTS
[171,4 ?Tim ki
Amor
PHS- 4667.2 (447)
FIGURE 16. Specimen Demand Search Formulation Record.
27
terms from the master MeSH list. However,other search elements such as author names,journal titles, and year of publication maybe used as well. Table 4 lists the search ele-ments which can be used in MEDLARS.
Table 4MEDLARS Search Elements
ElementSymbol
MS
G
T
z
C
A
ILQY
Name
Main HeadingSubheadingGeographic HeadingForm Heading (e.g., Review Article)Main Heading Used Only as "Print"
Term for Index MedicusMain Heading Used as non-Index Medi-
cus TermCategory Number from MeSH Hier-
archyAuthor NameJournal Title CodeComputer Entry DateLanguage AbbreviationPlace of PublicationYear of PublicationMain Heading/Subheading Combination
The number of elements used in any givensearch varies. The searcher also may "ex-plode" a subject heading by using the sym-bol "E" after the term and designating thenumber of levels in the MeSH hierarchy towhich the term is to be expanded. The com-puter will then search not only for the statedterm, but also for other terms specific to itin the hierarchy. In addition to the MeSHtree tructure, the searchers have developedinformal collections of headings which oftenare useful in search formulation as "hori-zontal" patterns of relationships (referred toas "hedges"). These hedges cut across theseveral MeSH categories and allow groupingtogether a collection of terms which identifya concept required in the search. Trees andhedges are an important feature of MED-LARS search formulation.
Groups of terms are frequently used to-gether in search formulation. These can bereferred to logically in the fomulation recordby use of the "sum" element. For example:
28
M1=EpilepsyM2=Epilepsy, grand malM3=Epilepsy, petit malM4=Sum M1 to .M3
After listing and grouping the search ele-ments, the searcher prepares from one tothree search statements. Individual elementsand sums of elements are linked togetherinto a search equation by use of three oper-ators : OR (-1-) ; AND (*) ; NOT (). TheGREATER THAN OR EQUAL TO ()and LESS THAN OR EQUAL TO (S.) op-erators are used with the computer entrydate and year of publication search ele-ments. The searcher has the option of for-mulating subsearches to provide the simul-taneous expression of as many as threelevels of increasing specificity (referred toas Sections 4, 5, and 6). The first statementmay be a broad expression of needs; thesecond may introduce limitations ; and thethird may impose the greatest specificityfor the query.
After completing all work on the searchformulation itself, the searcher next deter-mines the format for printing each citationin the final bibliography and the sequence ofcitations (e.g., alphabetic by author names ;alphabetic by Journal Title Abbreviation;expanded under subject headings) . He choosesfrom a number of format and sequence op-tions, and records his choices on a ReportGenerator Request Form (see Figure 17) .
He may also assign a title to the search. byadding this data to the form.
The Search Formulation Record and Re-port Generator Forms are then sent to theHead of the Search Section for review, priorto conversion to machine-readable input forthe computer.
In examining the role of the search special-ist, it should be noted that these highlytrained and experienced subject specialistsserve in an interpretive role between scien-tist users, indexers, MeSH, and computerfacilities. They must develop "hedges" to copewith rigidities of MeSH, develop strategiesto compensate for possible inconsistenciesor inadequacies of indexing, and develop rou-tines to cope with machine-processing prob-lems which occur. In the original systemplanning, inadequate consideration was given
REPORT GENERATOR REQUEST FORM
RCF Tape No.
Batch No.
RequestNumber
FormIden-tifier
No. ofcitations
Outputmedia
Trac-ings
Sort Keys
M
Headings
G H P Y
No.ofcols.
CitationArrangement
NJ
ColumnWidths
NJ
PageNumber
NameSectionPage
T Col 1 Col 2 Col 3 Col 1 Col 2 Col 3 Number
RJ RJ LJ ZLR LR
1-6 7-8 10-14 15 16 17-24 25 26 27 28 29 40 41-44 45-47 48-49 51-52 53-54 55-56 58-62 65-69
070852 F 1 ALL C X SA
RequestNumber
CardIdentifier
First Page Title - First Line
1-6 7-89-80
070852 H1 Molluscum contagiosum. 6- Venereal disease. 5-Etiology. 4-Other
RequestNumber
CardIdentifier
First Page Title - Second Line
1-6 7-8 9-80
070852 H2 aspects.
RequestNumber
CardIdentifier
First Page Title - Third Line
1-6 7-8 9-80
H3
RequestNumber
CardIdentifier Column 1 Heading/Column 2 Heading/Column 3 Heading
S
FIGURE 17.Specimen Report Generator Request Form,.
to the requirement for training these spe-cialized personnel, and training has occu-pied a major portion of the time of thesupervisors in the Search Section. The 'intel-
lectual interface between the user and MED-LARS has developed into one of the morechallenging and exciting aspects of the total
system.
Preparing the Search forComputer Input
Figure 18 depicts the flow of informa-tion through the Retrieval Subsystem. Thecreation of the Demand Search Formula-tion and Report Generator Request Formshas already been described. The next step isto prepare this information for computerprocessing.
A clerk in the Search Section matches theDemand Search Formulation to a tub fileof prepunched MeSH cards and pulls one
card for each subject heading used in thesearch. The purpose of this step is to cutdown on the amount of keypunching requiredfor each search, andwhat is perhaps moreimportantto reduce the possibility of cler-ical input errors and limit the number ofsearches rejected by the computer. The pre-punched MeSH cards, Demand Search For-mulation, and Report Generator RequestForms are then forwarded to the Office ofComputer and Engineering Services.
The keypunch operators prepare two decksof cards for each search : a search deck anda report generator deck. In creating thesearch deck, the keypuncher completes punch-ing of the prepunched subject heading cards,punches new cards for other search elements,and punches new cards for each search state-ment. The report generator deck includes oneformat card and up to three header cardswith title information for the final bibliog-raphy. Both decks are batched with other
29
MEDLARS RETRIEVAL SUBSYSTEM FLOW CHART
30
PRE-PUNCHED
MESH CARDS
+
\ANALYZEREQUEST
4
DEMAND
SEARCH FORMULATE
FORMULATION SEARCH
4.4.4%.....1 \,.t ---rt
FURTHER
CLARIFICATIONN EEDED?
NO
REVIEW
& REVISION
OF SEARCH
1
PULL MeSH
CARDS FOR
SEARCH
(KEYPUNCH
& VERIFYADDITIONALDATA
1
SEARCH
DECK
\BATCH FORNEXT
COMPUTER
RUN-
\PREPAREPRINT
SPECIFICATIONS
YES--1,- FURTHER
INTERACTION
WITH
REQU ESTER
/BIBLIOGRAPHICSERVICES DIVISION
OFFICE OFCOMPUTER & ENGINEERING
SERVICES
1
KEYPUNCH
& VERIFY(
IIREPORT
GENERATOR
DECK
1
\BATCH FOR
NEXT
COMPUTER
RUN
FIGURE 18.
6
MEDLARS RETRIEVAL SUBSYSTEM FLOW CHART
SEARCH
REQUESTS
ERROR LIST-
REJECTEDSEARCHES
INPUT &EDITING OFSEARCHES
MEDLARS
DICTIONARYTAPE SIGNIFICANT
ELEMENTS
LOGICALSEARCH
STATEMENTS
DEMAND SEARCH
MODULE
4
COMPLETE
LOGICALSEARCH I
FINAL
TAPEPREPARATION
1......7
I
FIGURE 18a.
\nRETURN TO
SEARCHERS
1
31
MEDLARS RETRIEVAL SUBSYSTEM FLOW CHART
IREPORTGENERATOR
CARDS
SORT INTO
FINAL
BIBLIOGRAPHY
SEQUENCE
LEXTRACT &
SORT MeSH
RECORDS
I
LOOK UP
Me SH NAMES
FROM CODES
& SORT
INTO FINAL
SEQU ENCE
f FINAL
BIBLIOGRAPHY -- *-PREPARATION
REPORT GENERATORMODULE
FIGURE 18b.
[-DEMAND
BIBLIOGRA-
PHIES
RETURN TO 'SEARCHERS
FOR SEARCHES
TO BE PRINTEDON GRACE
1'
;`;
ft
searches and forwarded to the computerroom for processing.
Searching the Computer Files(Demand Search Module)
Programs for searching the computer filesof citations are referred to as the DemandSearch Module. This module has four majorsegments or submodules (see Figures 18,18a, 18b).
The first submodule reads into the com-puter all search decks in the batch beingprocessed and performs a comprehensiveedit of each search. All search element sym-bols are validated, and the search statementsare checked to insure that only valid oper-ators have been used and that the searchequations follow the established rules forlinking elements and operators. Each subjectheading is then split into a separate recordalong with the search request number. Theserecords are sorted into alphabetic order andmatched to the MeSH file on magnetic tape;alphabetic MeSH names are converted totheir equivalent code numbers; and the fre-quency tallies of MeSH term usage are addedto the records. The MeSH records are thensorted back into original search number se-quence and re-matched to the search records.An average of about 7% to 10% of thesearches are rejected by the computer edits.
The next major task performed by thisfirst submodule is the creation of a table ofsignificant elements. A significant elementis defined as that element (or elements) ina search which must be present in orderfor the search to be satisfied and which hasa lower frequency of usage than other ele-ments "and'ed" with it in the search state-ment. An example will help to illustrate theselection of significant elements:
Search Equation:(A and B) or (C and D and E)
Frequency Tallies:A 629B 1016C 543D 816E -- 439
Significant Elements: A and E (i.e., ifneither A nor E is present in a citation, thecitation cannot qualify for the search).
Significant elements are selected for allsearches in the batch. Duplicate terms areeliminated and the resulting table of signifi-cant elements is stored in internal memoryof the computer for use in the next sub-module, High Speed Search. The significantelement table also includes a code for eachsearch in the batch to which this elementapplies.
The final task of the edit submodule is toformulate each search in the batch into alogical, decision-table format. These recordsare written onto magnetic tape for later usein the Logical Search Submodule. The use ofthe decision table will be explained later.
The High Speed Search Submodule is theonly Demand Search Program which re-quires passing the entire CCF File. TheMEDLARS computer system has overlapprocessing capability; that is, simultaneousreading of one magnetic tape, internal pro-cessing, and writing of another tape. In or-der to minimize the amount of tape-readingtime, it is desirable to have tape-read timeexceed internal processing time so that nostarting and stopping of the tape is required(since some time is lost in acceleration anddeceleration of the tape unit). The use of thesignificant element table in High SpeedSearch allows this program to process theentire MEDLARS CCF File in a minimumamount of time. The actual tape-processingtime for any given search run is a functionof the batch size of the searches beingprocessed.
The significant element table is searchedfor each citation record on the CCF, withcitations which match recorded on an inter-mediate output tape, along with the codesfrom the significant element table. Thesecodes indicate which of the searches in thebatch this citation could satisfy.
The next submodule is the Logical Search.This program reads in the decision tablesfrom magnetic tape and matches the cita-tions retrieved from High Speed 'Searchagainst these tables.
Note that each citation being searched inthe Logical Search Submodule carries a codeindicating which search(es) in the batch itcould possibly satisfy, and only the appro-priate decision tables for each search are
33
1
Table 5Decision Table Logic
*
Example: M1 * (M2 + M3 +M4)"and"
= "or"
RetrievalElement
No MatchCode
MatchCode
M1 5 0
M2 0 3
M3 0 3
M4 5 3
Match & No Match CodesO= Examine next element3=Retrieve this citation5=Reject this citationNo Match on M1
Match on M1No Match on M2
Match on M2No Match on M3
Match on M3No Match on M4
Match on M4
in table
Reject citationJump to look at M2Look at M3Retrieve citationLook at M4Retrieve citationReject citationRetrieve citation
examined. Those citations which satisfy thecomplete logical search criteria are writtenonto a tape of retrieved citations. Citationssatisfying more than one search in the batchare written a multiple number of times onthe retrieved file, with each citation identi-fied by a search request number. The LogicalSearch Submodule also produces a printedlisting of each search (and subsearch), show-ing the number of citations retrieved. Thislisting is returned to the search specialists.
The fourth submodule prepares the finaltape format for the retrieved citation file,sorting the citations into order by searchnumber. If the computer operator notices anunusually large number of citations re-trieved, he may check with the searcher be-fore printing this search, since the searchermay choose to print only a few citations andpossibly reformulate the search.
The MEDLARS two-phase search logicoptimizes the passing of a very large serialmagnetic tape file.
Printing the Retrieved Citations(Report Generator Module)
The function of the Report GeneratorModule is to print demand bibliographies in
34
a variety of formats and sequences (see Fig-ure 18). The module is called "Report Gen-erator" because the computer generates anumber of different possible report formatsbased upon the specifications given to it bythe report generator cards (specified by thesearcher).
The first program of this module edits thereport generator cards for errors andmatches these cards to the retrieved cita-tions for each search in the batch. Thosesearches requiring printing of citations bysubject heading are expanded ; that is, thecitations are copied once for each subjectheading to be used in the expansion.
The resulting intermediate work tape nextis sorted into final sequence, using the sortkeys provided by the report generator cards.
If any of the searches in the batch re-quires printing of "tracings" (a listing ofeach subject heading with the printed cita-tions), a third submodule sorts the subjectheadings into numeric sequence and looks upthe alphabetic names on the MeSH file. Theseare then re-sorted back into search requestsequence.
The final submodule creates a magnetictape for printing of the demand bibliogra-phies on the Honeywell-200 computer. Thesebibliographies are returned to the Search Sec-tion for screening and preliminary evalua-tion, prior to mailing to the requester. Thesearchers employ several form letters to ex-plain the results to the requester. In addition,each search is accompanied by an appraisalform requesting a formal evaluation of thesearch by the user (see Figure 19).
As of January 1, 1968, the average turn-around time for searches (receipt in the Li-brary to mailing of results) was about twoweeks. Throughput time varies substantiallywith the workload, however.
It is also possible to print a demand searchon GRACE in those few cases where thedemand bibliography is to be widely dis-tributed to a large number of people. In sucha case, the Report Generator Module pro-duces a tape which is processed by the OutputModule, described in the next chapter.
No. of Search Name of Requester
The National Library of Medicine
APPRAISAL
OF
MEDLARS SEARCH
BOB No.68R 891App. Exp. 6/30/68
Date of Request
Your frank appraisal will greatly assist us in evaluating and improving MEDLARS
(Medica). Literature Analysis and Retrieval System). Thus we hope that you will
complete this form at your earliest convenience and return it to NLM for our
review.
Joseph Leiter, Ph.D., AssociateDirector for Intramural Programs
Relevance
How many citations received were relevant to your request?
In your opinion, what may have caused retrieval of the non-relevant citations?
a) indexing / /
b) search formulation I /
c) subject headings in Medical Subject Headings / /
d) statement of request L__/
e) other 1 / Explain
Completeness
In many cases it will not be possible to determine the completeness of the search
for periodical articles. However, it would be helpful to know how many pertinent
citations known to you were not included.
Would you please list at least the more important: ones?
Timeliness
Did you receive the citations in time *to be of use ?`
Do you have comments on timeliness?
FIGURE 19.Specimen Form, Appraisal of MEDLARS Search.
(Continued on Page 30
35
General Comments,
General comments not covered above may also be valuable in improving this new ex-
perimental service. These comments may be grouped as follows:
a) In your opinion, are Medical Subject Headings (found in Part 2 of Vol. 8, No. 1,
January 1967, Index Medicus)
adequate? L./
inadequate? L../
If inadequate, do you have comments as to how they may be improved?
b) In your opinion, is the List of Journals Covered for Index Medicus (There is
a list in the January 1967 issue of Index Medicus)
adequate? i__/
inadequate?
If inadequate, do you have suggestions as to improved coverage?
c) Other
FIGURE 19.-- Continued.
Chapter V
PUBLICATION SUBSYSTEM
The Publication Subsystem consists of pro-grams and procedures for producing MED-LARS bibliographic publications. A completelist of these publications was given in Chap-ter II, Table 1. This chapter will .discuss themethods followed in preparation of recurringbibliographies and other MEDLARS publi-cations.
Figures 20, 20a, and 20b are flow chartsfor the Publication Subsystem which is ref-erenced throughout this chapter.
Establishing and TestingRecurring Bibliographies
A recurring bibliography is a periodic se-lection of citations from current input, struc-tured by a predetermined pattern accordingto the interest profile of a group. After beingcontacted by a professional organization witha request for a new recurring bibliography,Library management reviews the request inrelation to available resources and makes adecision whether or not to proceed with thenew publication. If a decision is made to pro-ceed, several meetings are held with repre-sentatives of the cooperating organization toselect tentative subject and format parame-ters for the new bibliography. This selectionwork is coordinated by a senior staff mem-ber of the Search Section, BibliographicServices Division.
The tentative parameters are tested byrunning several computer searches to pro-duce "test" copies of the new publication. Theresults are reviewed jointly by NLM and thecooperating organization, modifications aremade, and the search repeated. This testingmay require several months' time, dependingupon the complexity of the subject. In somecases, the test bibliographies are mailed to a
select group of outside experts for reviewbefore the final parameters are selected.
After the test search has been approved,the final specifications are forwarded to theInput Unit, where recurring bibliographysearch and format specification cards arekeypunched and verified. The search specifi-cation cards, which are punched in the for-mat required for the Demand Search Module,are filed and held until time to run the recur-ring bibliography. The format specificationsare added to the recurring bibliography for-mat parameter file.
Computer Programs for ProducingMEDLARS Publications
The recurring bibliography search cardsare pulled from the file when the publicationis scheduled to be run. (Note that the fre-quency of publication varies from monthlyto annual, depending upon the recurring bib-liography in question.)
The search cards are run through the De-mand Search Module (see Chapter IV) andthe citations which satisfy the search arewritten onto tapethe Retrieved CitationFile (RCF). Note that in running the De-mand Search Module, only the "current" CCFtapes are searched; that is, those new cita-tions added to the file since the last run of thepublication. This is controlled by the datecard input as part of the search.
The next computer, job is running the Re-curring Bibliography Submodule (see Figure20). This submodule processes the RCF tapefor the recurring bibliographies and formatseach citation on the RCF into standard printformats. Each character of the citation isconverted to a special code required forGRACE, and the citation elements are ar-ranged in the correct format for printing. A
37
38
MEDLARS PUBLICATION SUBSYSTEM FLOW CHART
DECISION
TO BEGIN
NEW
PUBLICATION
B!BLIOGRAPHICSERVICES DIVISION
COMPUTER &ENGINEERINGSERVICES DIVISION
DEVELOP
SUBJECT &
FORMAT
PARAMETERS
I
RUN COMPUTER
SEARCHES TO
TEST THESE
PARAMETERS
YES
\PREPAREFINALSPECIFICATIONS
FOR BIBLIOGRAPHY
RECURRING
BIBLIOGRAPHY
SEARCH SPECS.
)KEYPUNCH
& VERIFY
APPROPRIATE
DATE CARDS
FOR SEARCH
DEMAND
SEARCH
MODULE
(SEE CH, IV)
(USED ONLY WHEN
INDEX MEDICUS IS
BEING RUN ALLCITATIONS FOR CURRENT
MONTH ARE USED
RECURRING
BIBLIOGRAPHY
SUBMODULE
PROCESSED
CITATIONFILE(PCF)
0FIGURE 20.
DATE
CUTOFF
PARAMETERS
(INDEX MEDICUS
ONLY.)
MEDLARS PUBLICATION SUBSYSTEM FLOW CHART
EXPAND BYMEDICAL
SUBJECTHEADINGS
SORT INTO
PUBLICATIONSEQUENCE
OFFICE OFCOMPUTER & ENGINEERING
SERVICES
FORMAT
PAGES OF
PUBLICATION
ITP1.--'E(S1FOR
GRAPHIC ARTSCOMPOSING
EQPT.(GRACE)
[ GRACE
ELECTRONIC
UNIT
w=4.- GRACE
OPTICALUNIT
EXPOSED
PAPERFILM OR R MOM
REFERENCE
SERVICES
DIVISION
INSPECT,
CUT,
PACKAGE
J
AUTOMATIC
FILM PAPERPROCESSOR
MAIL TO PRINTER(OR SPONSORING ORGANIZATION)
FOR PLATEMAKING,
PRINTING AND BINDING
PRINTED
RECURRING
BIBLIOGRAPHY
FIGURE 20a.
f YES
DEVELOPEDFILM ROR
PAPE
39
ILIST OF
JOURNALS
INDEXED
SUBMODULE
,
/
MEDLARS PUBLICATION SUBSYSTEM FLOW CHART
'From Medlars Dictionary Tape
BIBLIOGRAPHYOF MEDICALREVIEWS.CUMULATEDINDEX MEDICUSSUBMODULE
LPUBLISHED
MEDICALSUBJECT
HEADINGS
SUBMODULE
FIGURE 20b.
partial right-margin justification is also per-formed by varying the amount of blank spacebetween words of the citation. The resultingformatted citations are written onto anothermagnetic tape called the Processed CitationFile or PCF.
The Recurring Bibliography Submodulealso is used to create a PCF tape for themonthly Index Medicus. Since almost all newcitations entering the system go into IndexMedicus, it is not necessary to run the De-mand Search Module for this. publication.Rather, the new CCF file for the month isinput directly into this submodule with theappropriate date cards.
The programs which process the PCF andproduce the final tapes for printing are calledthe Output Module (see Figure 20).
A submodule expands each citation by sub-ject heading, creating an Expanded Pro-cessed Citation File (EPCF). The EPCF isthen sorted into final bibliography sequence,and 'a Format Submodule produces a tape forthe PhotocomposerGRACE. Note that inthe case of Index Medicus, the twelve monthlyEPCF tapes are saved for the annual camu-lation.
The formatting programs are quite com-plex, since they must deal with various for-mats (1, 2, or 3 columns per page, etc.). Inaddition, the programs handle such things aspage headings, intermediate headings withincolumns, and continuous page numbers (seeFigure 21).
The Output Module includes three othersubmodules for producing four special MED-LARS publications. The BMR/CIM Sub-module takes the twelve monthly EPCF files(which have been collated by subject 'and -au-thor) and produces two sets of tapesonefor the Cumulated Index Medicus and an-other for the annual Bibliography of MedicalReviews. These tapes are processed by theFormat Submodule to create tapes forGRACE. The LJI (List of Journals Indexed)Submoduls uses the Journal Record File(JRF), created by the Input Module (seeChapter III), and creates a tape for GRACE.Similarly, the MeSH Submodule creates aGRACE tape for the annual printed list ofsubject headings, working from the masterMeSH file.
The MEDLARS Photocomposer(GRACE)
Tapes produced by the Output Module areprocessed by the Photon 900 computer pho-totypesetterGRACE.
GRACE consists of three components.Print lines are delivered from a standardHoneywell magnetic tape drive. The elec-tronic control unit of GRACE includes an op-erator control panel, an input converter, asmall amount of core storage, and special-purpose logical circuitry which computestiming signals for driving the optical unit.The optical unit has five major subassem-blies: flash tubes, a matrix of charactersetched onto glass plates, a reciprocating lens,a mirror block assembly, and film handlingunit. The lens makes one complete horizontalsweep for each print line, during which timethe control unit sends a signal to each of theflash tubes at the exact instant required toexpose the appropriate character at its prop-er position on the film. As the lens sweep iscompleted, the film is advanced, a new recordis read, and the cycle is repeated (see Sche-matic Diagram, Figure 22).
The final product produced by the opticalunit of GRACE is a roll of exposed film orpaper, full size, 9" wide by approximately120' long. Each roll holds about 120 pages forpublication.*
Final ProcessingAfter the tapes have been run through
GRACE, the exposed film or paper is devel-oped by an automatic processera KodakVersamat.
The developed film or paper then is sentto a quality control station where it is in-spected, cut into page-size sheets, and pack-aged for delivery to a commercial printer.Offset printing and binding completes thePublication cycle.
For a MEDLARS recurring bibliography,the above cycle is run twice. A galley proof
For a more complete description of GRACE see:Paul C. Redin r, Publication by GRACE, Methods ofInformation in Medicine, 4:145-50, September 1965.
41
Ir
4
z.
4'-v
4
IntermediateHeadings
42
Page Heading
INDEX MEDICUS ARACHNOID
AHATOMY H!STOLOGY IPd.C1,IU.lt.t..I...I.fbdI,lHKkr.!.M..t 1Th .ft.d SI $I$I1$ Id IflIa!Utl .$ I.J M.d MK 3I%41. SU H (Gs) c..U. .1 ft. ki .f th.
A Mt.UtlJ ty .1 t. .pp.$dl .i.U.i.1.. A Itb. N. t L_l. ti Idj l4ely ! nIØJ VP4t4! P4!! SpC KSI 191 Spu 41 WL I At 13 (1*)l&D41. !dZl M. t J.D.9p M.d C%U4 NAI tIS.II. Fb IT iIItl.. d.t.dI Id ld..UflcUS. a l.. .1 D*VO EFTZCI1
VIZYMOLOGYd dng.. .1. 1 lc y b.
Nd t. Gdlly I. .pA.i. I.t.0 AH. .t Tb. .. SS..dyt.!. tb. t,.Ia..t .1 IU.I_ J Pka PkMU! !t37344. M.y IT 11d C tflIC Prt M l.clly .1
t4tk d.bY4NIUK rn t. l.fl..4 , ,?U9 . ,.p.:.4 ti thit SI $I.,p*..,"&l. p1cw L.t i. I.' $.g 47 43I.. , $7 APPETITE DISORDERS (C4. F)
!NNr1VATON T)rIAPY Atly.4b..4d .q.I'J b.wIn..a.UbWUl fl ct I.tm.c,: tIIu. Ml b1
A bI&NIUI ty 1 tb. ,n.dI .i.U.t1. A (Ttt .1 .14 .pt!t. k.a ! ctiId.iI tUflcUly. u4 P. .t II,...lw. I. l4ely l rSiØJ D.b.7 P. lIt $t 44 13-I. Jl ITdlUH! !dkItl M. .1 J C.0 M.d N.p p.n. S 834. O J. ST (Fr)De.Ip M.l U4 N...I tN4Il. Fô ST M!TAO13M
APPETITE (F)VITIUMAY
Ca.S.. . &..I (lE Il Mt. g!ta.k Istd1_ th. k1II.. ettt Ii tb. g.w IV. HhaUI.. c,.ts I. tb. bMt an.,,J
11 ..ICUC ethlly CS f a pt*flII bCtIil IsU.j KpV$dMI4 _2 p!II l S$U CA. t J. .ydflJ l lupflt*. a p11.. WlIzl DA. .1 .L Wi. t iL kB 3$ 3ZT )O ct4. (p.1)
A.. J PT.9l 2!3 3t7 2. AI IT C.,..0 V.t T 7$. Jn ITA,JI. I . (44 St...a.. PNYSIOtOOYJA..I J C.4J py9l PU*l4 54-4. Oy ST APRAXIA 'ClO ')*.Kt!.0 .dl.J 4 I.t.I.1I.n. .1 th yflIk1 CtI4 1t lb4 DW s.d c.utcV..I lysp.I. It.L 4 tr t mbWI 7 .jt. c.n1I t.fl...da., 1lIk IA. .1 J. kk.c. 97 UT4 $. 2$ S. IT W.CUI.f VbI 1.4 IQ. I. .I,dt.. ,.U.. I. t M 4 h.. th.lI..IldI. l4 .!.t:l.. 4l..-gdtlS DI. U J. .t .J A.. J M..I D.Ic TI 04-1. J.. IT t1L $.. ML .t Ll.cI.I H.!! F. I.i. prkt.I ZI mat O, ITMa.. M.d W.c IN $5-fl. 14 H (Or) APTITUDE (F)
tXVG r?ItCT3 A1 llll.et$ Is tI!d 4 cIlIfld Id.lZ..c,Hw JL .t ii Act p.1b.l (AI) )S !7-21. lIST pI.b..5..It.sr .tS1.I risds.7. ..ul1.lC .lC.ct. Cl l, l M.t.l utt.t .1 !Ci . II4s.i .vht.4 1W 5tnafltfl P. .t IL A. J OØL! 14 11!4. $. ITII.d *lt .Ibl?5le$. GJ. 1 l. l.lg flI..111. HIdI JT. .! !
c...l J PT.9I Ph$C! 4 544T. M.y ST A.. 3 M.t D.Ik TI M.y H ARABINOSE DIICP*4 VI. S..,. A eer..tf pt$C.dT. Ut MICIIS k1tØu.th7llI*. c.t.. L.A. .t l .4 .kw clId.. is KI WD N.cI.sI4.. XXXV.J sd S 1400S. Dee NPhu..HlC IdI.. 1tI
Eu,U..aI Cl!4 33 235-4!. Dee 14Sf t. ..Ise. th
I .I. DsTIL t 1 J M.d Cb IP)4T4. ST
S p.dIetI.. .1 HHI1I R. .t H4I .1 (.I ,I.tfl.I!).4.1. lidV.11k..., W eI .J. J Id 54 $9.33. JI 4T ! J A,,. P.yc.l l 5044. F. IT ((4 D.flb!..I 7).flJ. P.4t !I. t I!
Ip...es1.glc 4y .1 .UI1.tC S.It UI l b10fl J M.d !I!). .1.. 4T
II..n.l JR. I !. tI p .t.dbl MCI.$.14. V. 2t3*pt.W!i(An ml IST.ISO-U. M.y ST (Fr) AK. J A$ PTcl SI liT. F. ST b.t ibj. WV. .1 ILVtl.*l!.l..IC*pH*.tI Pmy J M.d CI 1133!4. O7 IT
PPY$IOPATHOLOGY DX. I ! J App PTCI $1 1U34. F.b STCM,.d.,1IUcI .1 pU t tI 1 eI!.Uy. M!TAOU$M
Hyp.tIlC I..I.. ,11.d I. IM..tti .1 U I. I I. J Cfl. P.ycb.I 33 ilaO. J. 4TU.I,. dd dIet. Mfl1 U. t 1 CpI I,.t,g ! .l!lS t s.d M.bthly .1 tb. lu.. s.d .1 tb. I..4.d ......c.fld.
J Py9! 213 $4-5. S.p IT cII..d L4.d.0 .4 st4.,IC .pUd.. L4 F! 1 &Ct .I$ ktit A.Pyc.l Rp )O US-U. F.b IT II.p. At 133 113 . F.b IT
APPETITE DEPRESSANTS (1)5)bn ..t.b. !i HMrn...
QflIII. ..4 t..1 l* Pyc! z O 3!S-21. Fib 4T J MKrNl U YZ . J IT
chkIrII. y4KlnI4. S.iLa IF .t i, SI4l . th .ltI..3fl ..0 I p.,f1e1.c. Frtb. 1w psU. e.tril .1 I
J $d 54 101-3. AZ ST L.rnW. .$U.b, U...nC lb.ppajd DL .t II.
pd.dp. ICt4 Itb. L 1 l P.Tt.I Yr$t 3t2!l.4. PS4 (Gu) J M.c 11.1 3 443.54. 5 MU St
r.t,. . Ig.I 3.flW L.a.CkIl $pdfld!y .1 ffl! Tytht*7tM IMk. L.Kdk.fI.. MIkI A. .1 APTITUDE TESTS (F) MIfl (L..I..) 3!I 5134. 4 F.b 4TCR Ae4 d (DI (Pad) 2 l34-L I Mt IT (?)
Tb. MIdc. (IllS) 4 t. DIl.s ARACHIDONIC ftCID 'DIIADV1$L IFFICI1 1044) 1U I c...calml t.y. .fl AA.
mt J !4.I P7tbI 3T.IIT. F.b 4T M!TAOL*$MW.I1t pW. RC. H..dIIy !.tI..C.. . itI!JAMA 2$Z E4. I! £.p IT mby .c.t.. .1 MIrn.I VU.l
J...t*,y. V..d.b.fg 5G. t !.H.t.l4Z..It7I lb. c..p..IU.. .1 et.I.SL.I.I ..t...ylb..U.d by s!l trsetl... .1 lu(.d 114 1.4
PNAIMACOOYMAMICI J Ap,I PTc.l ! 11t F.b 47AuT Si.1 I4 .HUI $ptIlL G.vwlI ZI.
i.1 S.0 L t ILII..a IIØ).1 1131*-Il. 13 Ja IT
7. tlKt .1 AaIW.. P.ycl 13 $S-I. 4T ru9 gr.dk. n...l Ip4 ...U, r...,. *14YhI.ky J. *1 .J, mCJ.IC d4 SaiCt..*l lt!.S-
An t Prk4y 14 "1*. Dee N rT. .t .1. MII M.d 132 44. J.. 4T R..b.Z A. 9?III. S $.p ITA .1 t. .IC. SI T..t .14 ..t.d ,tI.. .1di.Uippl....4 Mntflt! .n c,hlIJ tIlct I.114.l 11c ill 'ARACHNIDISM (C14)IdI J p.1 O )O?.II. J ST Tb. I.fl. - f I Sky *p4.t. lt 1 ..d.L
snf.t In a cs4 St.. .... .Wty M, Pycb!1p 30M ?.bdT J M.d AN Ab.a * 1N. MU 4TJA..t II C...4 J PysI.l P..- . . .l.tl.fltM ..0 S p.f..Ise.
' ARACHNODACTYLY (C3. CI;)._.... .1 t i
... J M.Ø....I I23344. JZ IT S.. fltb!41& sap.CU .1 .t.dlta, dINi4e,S .1
t U. (Uf&' .lr.iRtIII .1 id !Ik I fltUI S. VIKt AQUEOUS HUMOR 'A9 AIZ' CdIy%.r.i)I G3I H. t IL4,I,dM.r.d !IU.ru.lMil, u4
!nlV..u.IIy t. ..p d £..I. *g4t IL Tb. .11Ct .1 L c.fl p4 t. flWNS!' P.d111t (Iu.) 2$ 14111. H.,'D.c N ()J D.Jy 3d1 $OIITS*. J! 4T p...*bIlfly 1 t bfl1.
St., 6. .t .J. TAIOUAI. a OIWIC.J J ph.&c.I 1t33T.$. M.yIT M.d Pb*l rp (au.l) lI14T !. INTr.r.es1ulc tdl.. 1t Th..11.cU.lI.lUl.l.tntl...fg,Ia.tId!..$M (The z.s.Uc ld9I a arfu. y%lN3 pb.y!p py)t.I4I. 1W 447 CN.. l .(... C,.p$.1U In (Ob..fflU.S .1 ltatb. a. 11.111!. lt.I
W. l .J J Ph.. $d 54121-33. Jl.IT bbIt. DZr P. .t II. £.mU.U)J a., C.
Ip.n..e.l.lc .t.4, C ,uI1..lC .1..I O,I.I.l.IIca (I..l) 133 )Sl-2. ZNT T.,ta ..nIuMt. It ISO.. Ju. N (O*)l.rfrnttI Icu1. JR. 1 J. Tb. sIt,ut,.di .1 t. .pU. .1 t. cthu,AnI l Ptrk..4y. 1ST. 1S4-U. M.y IT (Yr) b,dy. A tdy .1t. I,IcUI.lespIm. .4.1 tb. ?A?OLOOYUpl4 .bUlM etI. .1 ... $ft.fUlC rug.J CUS IC1at4 .111 tb. p,$d.cV. .1 ph44
.cI! L. .t .J.1.0 $.0 !tl 191 $pt 41 !4fl4. 3% Dee U (lU
sq.... Iltl A h. .1 I.Z 2.UftNk 5t4354& 1114 Mufu sdrJ 5tk1g HH. t .L
T,sMt.ii Z ptt 7I$l.U. lIST (O.)
TII!Ap!LI1C U! AHALtSARACHNOID (As)F.fl,.II.lI t. rnl.I 1 bUUy 3S AW..ti Iat J CIl p,ut DeeN
(l.a. ..I.dr*pbw.tIC 17d4 f
.q..oa ba.nJ Fur. JP. t-.J. 3.I.,.U.1sI.,t..Mal..lq.t4..I.tfla$:A 00(41 (P*l1I) 1(773.1. Dee N (Tv) . pIct. ?Ut . :.a T4ll.L UI, ST
Part 2o1 the JL1iiry lue contt s.ciegored It 01 subject headings. ndtln$ crorefercnc'u.ThIrIOCItIOn 101in the categorized It U IndlcsIed by the le er.irn,cr &lgivion (M. 8), etc.), fC4Iowlng each ubect h.dlng.
FIGURE 21.-Sample INDEX MEDICUS Page.
RunningHead
MarginJustification
ContinuousPageNumber
CONTROL PANEL
CO CM13 0:0 0r a iiiii
(SEARCH
WOE
SEARCHCOMPARATOR
TAN RawTRANSLATOR,RARrtYCHEcx61A 0:HTIKA.
CADEDETECTOR
TAPETRANSPORT
Doman catsLC*0 TABLE
MODECONTROL.
INDEX WOE
TRANSCRISE MODE
RANK and
CALLuanCHLOGIC
WIDTH
1.4E44ov-2048 w:A20522 PITS / WORD
CHARAATCR ;COMFYCOVE And FLASH
POSITION1024 WORDS
SORTLOGIC
RANK and WIDTHTAIDLES
512 WORDS
LEADER IWO WIDTHTATTLE
4 WORDS
LEADER COSHSTORAGE.
128 WORDS
1-11.
Figure 5
Photo-Typesetter
Block Diagram
HORISCNTAL &VERTICAL RIGE SITE
2 160105
SPARE318 WORDS
CHARACTERREGISTER
LENS POSIToPULSE
COUNTER
CHARACTER CODE
CHARACTER ETI706E
CORESELECTMATRIX
264 LINES264 FLASHTuDES
AmPLJFIERL
II
LEADINGCONIRCt.
PHOT36OPHICCONT;CL
LEAC*4 MDR
LEADINGERROR
DETECTOR
ERRORFLASH
DETECTOR
CHARACTER =rmlMATRIX
OPTICAL FILMSYSTEM MAGAZINE v=::9
FIGURE 22.Schematic Diagram of GRACE Optical Unit.
copy is run first and sent to the Senior Sci-entific Editor, who checks it for scientificaccuracy (of indexing) and for any obviousmechanical errors. Corrections, if necessary,are made via File Maintenance Programs andthe final copy for the publication is pro-duced. Because of its size, this procedure isnot followed for Index Medicus.
Index Medicus, NLM Current Catalog, Me-dical Subject Headings, List of Journals In-dexed, Monthly Bibliography of Medical Re-views, and Cumulated Index Medicus aredistributed by the Superintendent of Docu-ments, U.S. Government Printing Office; re-curring bibliographies are distributed by thecooperating organizations.
43
Chapter VI
UTILITY PROGRAMS
Chapters III, IV, and V have describedthe main operating computer programs ofMEDLARS. In addition, there are severalUtility Modules which support the operatingprograms and which are quite important.These utility programs consist of three mod-ules and a miscellaneous category, as de-scribed below.
Updating the Subject Headings File(MeSH Generator Module)
The Medical Subject Headings File(MeSH), described in detail in Chapter III, ismaintained and updated by the professionalstaff of the MeSH Section, Bibliographic Serv-ices Division. This staff of experienced lexico-graphers anu clerical assistants produces ad-ditional and changes to MeSH for updatingthe magnetic tape file in the computer system.
MeSH is revised once a year and printedin a separate volume accompanying the Jan-uary issue of Index Medicus. A considerableamount of time is spent reviewing and se-lecting each new heading for the annualrevision. The subject heading specialists pro-duce a New Heading Work Sheet for everynew term entering the system.
This work sheet provides the new alpha-betic MeSH name, the definition and scopenote for indexing, information on authorityfor the new term, how this concept was pre-viously indexed, place in the tree structurefor the term, relation to cross references, andfile maintenance required of the CCF (ifany). In addition to new subject headings,the subject heading specialists also produceother changes to the MeSII file, includingheading and cross-reference changes, andwithdrawing old headings.
A clerical assistant in the MeSH Section
assigns to the terminology to be changed thenecessary code numbers required for thecomputer and produces the source documentsfor keypunching. Figure 23 depicts the flowof information for MeSH maintenance.
The MeSH Generator Module matches theMeSH change cards (which can include newterms, changes, and deletion of old terms)to the MeSH file on magnetic tape and makesthe appropriate changes to the file. Thismodule is run throughout the year to addand delete provisional headings. At the end ofeach year, the module is run to apply all thechanges created for the annual revision. Themodule also is used to produce the annualMeSH file which includes cross-referencesfor Cumulated Index Medicus and the annualBibliography of Medical Reviews (see Chap-ter V). The final function of the MeSH Gen-erator Module is preparation of printedMeSH lists used internally by NLM staff,including lists showing the MeSH hierarchi-cal tree structure.
Maintaining the Compressed CitationFile (File Maintenance Module)
Many situations create the requirement forthe ability to make changes to the masterfile of citations (CCF). Errors in citationsalready on tape are frequently detected byN'LM staff, users of Index Medicus and re-eurring bibliographies, and recipients of de-mand searches. In addition, the annual revi-sion to MeSH requires that corrections bemade to the CCF in cases where a one-for-onesubstitution of terms or a deletion of termshas occurred. (It is not possible to correctthe CCF when a general subject heading isreplaced by several more specific terms;MeSH is therefor revised only once a year,
*45
46
\
MeSII GENERATOR FLOW CHART
AP LR OAFNEASLS yl OsNi
\--o- V/ORK SHEETS -----0-LMeSH
CHANGES
COOING OF
OF MeSH
CHANGE
MeSH
CHANGE CARDS
UPDATED
MEDLARS
DICTIONARYTAPE
MeSH
GEN ERATOR
MODULE
FIGURE 23.
UPDATED
MASTER
MeSH
FILE
\
FILE MAINTENANCE FLOW CHART
I
FILEMAINTENANCEMODULE
FIGURE 24.
47
since the searcher must consider these gen-eral-speciE MeSH changes in formulating asearch.)
File maintenance changes to the CCF origi-nate in the Index Section, Search Section,and MeSH Section. Figure 24 shows the flowof information for file maintenance.
After keypunching and verification, thechange cards are processed by the File Main-tenance Module which corrects the appropri-ate records on the CCF. Note also that theExpanded Processed Citation Files (EPCF)for the twelve monthly issues of Index Medi-cus, which are saved and cumulated for theOM (see Chapter V), are also corrected bythe File Maintenance Module.
Maintenance of the entire CCF file (25reels as of January 1, 1968) is a time-con-suming job. This is normally done only oncea year, following the annual MeSH revision.
Reports Of System Activity(Statistical Module)
The Statistical Module produces manage-
ment reports for the Library staff operatingMEDLARS. Table 6 shows the reports whichare available from this module.
The module is run monthly, quarterly, andon an annual cycle to produce the reportsdescribed above. Figure 25 shows the struc-ture of the Statistical Module Programs. Aparameter card is punched, specifying whichof the eight statistical reports are to beproduced and what period of time is to becovered. These parameters are matchedagainst the Meal file and the appropriateCCF tapes to produce the printed reports.
Miscellaneous Utility ProgramsIn addition to the three modules described
above, other utility programs are used inthe MEDLARS system as needed. These in-clude such routines as generalized sort andmerge programs provided by the computermanufacturer, programs to copy magnetictape files, i.e., the CCF and MeSH files foruse at decentralized MEDLARS Stations (asdescribed in Chapter VII).
Table 6Statistical Module Reports
Report Name Description
Journal Report, List of journals indexed during the period, with date of publication of each.
Main Heading/Sub- Frequency of main heading/subheading usage with a breakdown as to Index Medicusheading Usage (IM) or non-IM.
Language Report Total number of articles in English indexed; total number of foreign articles indexedwith breakdown by language.
Index Medicus Number of citations in Subject ,Section, Name Section, and number of junior author
Statistics cross-references in Index Medicus.
BMR Statistics Number of articles indexed for Bibliography of Medical Reviews.
Category Usage Number of times each category of terms in the MeSH tree structure is used.
Indexer Statistics Average depth of indexing by each member of the indexing staff.
High VolumeHeadings
List of large volume main headings and most frequently used provisional headings.
48
s5
Cr)
1.5
Chapter VII
OPERATING EXPERIENCE: 1963 - 1967
The preceding chapters have describedMEDLARS as it was operating on January1, 1968. This chapter presents actual operat-ing experience during the period 1963 to1967, points out differences between the cur-rent system and the original design, andcomments on how well the original systemobjectives (listed in Chapter I) have beenmet.
Data Conversion PeriodThe initial plan for implementation of
MEDLARS called for a one-year data con-version period during the year immediatelypreceding production (1963). The plan wasto build a one-year file of citations on theCCF which would serve two purposes: (1)provide a large data base for final systemtesting; and (2) provide an initial file ofapproximately 150,000 citations so thatsearching could begin immediately after pro-duction.
The conversion plan required operation ofthe new MEDLARS Input Subsystem in par-allel with the Listomatic System for produc-tion of Index Medicus. Three special Flexo-writers (paper tape typewriters) were pur-chased to aid the conversion efforts. Papertapes produced for the Listomatic Systemon Justowriter machines were read and cop-ied on the conversion typewriters, with theadditional information required for MED-LARS entered from the keyboard.
The conversion plan met with very limitedsuccess. Many problems were encountered,the most serious of which are listed below:
(1) Because of delays in computer pro-gramming and preparation of the ini-tial MeSH file, the Input Subsystemwas not ready for use until April
1963, three months later than planned.(2) The abnormally high workload created
by the need to run para,A2,1 systems,coupled with the pressures involved infinal testing of the new system, forcedthe Library to abandon plans for com-plete conversion and to include onlyEnglish-language articles in the newsystem. This fact, along with the delaydiscussed above, resulted in only45,000 citations being converted totape during 1963, rather than theoriginal plan for 150,000.
(3) Many "bugs" and ..iefficiencies in thenew Input Subsystem had to be ironedout during the conversion period. In-dexing errors occurred because of lackof experience of the indexers with thenew system and the requirement for amuch greater depth of indexing.
(4) Finally, the deficiencies in the initialMeSH file began to become evident asthe conversion and testing period pro-gressed.
Despite these problems, two reels of CCFtape were created and proved invaluable asa resource for final volume testing of the sys-tem. These tapes were also used initially forsearching. However, the Library decided in1965 to discontinue searching the conversionfile because of indexing inconsistencies andMeSH problems, and these tapes are nolonger included in the MEDLARS data base.
Problems with GRACEGRACE had been scheduled for delivery to
NLM in May 1963, well before the opera-tional target date of January 1964. Slippagein the delivery schedule required Librarymanagement, late in the summer of 1963, to
Index Medicus(NEW VIEWS ON THE EVALUATION OF ACOU'IC TRAUMA)AUINGER J. MSCHR OHRENHEILK 97:68.U. FEB 63 (GER)
ACRIDINES D. D8I
STUDIES OF COMPOUNDS THAT INHIBIT CELL GROWTH. BALISME, ET AL. EXP CELL RES EUPPL9:472.8, 1963TREATMENT OF GENITAL TRICHOMONIASIS IN BULLS. FITZGERALDPR. ET J AMER VET MED ASS 143:259-62, 1 AUG 63
ACRODERMATITIS (Cl. C121
(INTERNATIONAL INCIDENCE OF ACPODERMATITIS CHRONICAATROPHICANS) DANDA J.HAUTARZT 14:337-40, AUG 63 (GER)
ACROMEGALY (C7I
ACROMEGALY. AN ACCOUNT OF TEN CASES. SINGH H. ET AL.J INDIAN MED ASS 41:109-14, 1 AUG 63
ACRYLIC RESINS (D8)
A PLASTIC EMBEDDING MEDIUM FOR THU: SECTIONING INLIGHT MICROSCOPY. CATHEY WJ.STAIN TECHN 38:213.6, JUL 63(RIGOLAC-METHACRYLATE EMBEDDING FOR ULTRASTRUCT'JRERESEARCH) FUJITA T. ET AL.ACTA BERM (KYOTO( 58:41-6. JAN 63 (JAM
AcTINOMYCES (B31
UNMINERALIZED FOSSIL BACTERIA. BRADLEY WH.SCIENCE 141:919-21, 6 SEP 63
ACTINOMYCETES (B3)
ANTIMICROBIAL ACTION OF NITROGEN.CONTAINING STEROIDS.SMITH RF, ET AL. J BACT 85:1295.9, JUN 63
ACTINOMYCIN (D3I
CLINICAL EXPERIENCE WITH "ACTINOMYCIN D" ON WILMS'TUMOR. HOME S. ET AL.KYUSHU J MED SCI 14:61-6, APR 63ACTINOMYCIN D EFFECTS IN FROG EMBRYOS: EVIDENCE FORSEQUENTIAL SYNTHESIS OF DNA-DEPENDENT RNA.FLIEINGER RA. SCIENCE 141:1063.4, 13 SEP 63
ACTINOMCOSIS (C11
A CRITICAL SURVEY OF MEDICAL MYCOLOGY FOR THE YEARS1946.-L956 IN JAPAN. TAKAHASHI Y.MYCOPATHOLOGIA 19:105-24, 20 MAY 63
ACUTE RENAL FAILURE (C6I
RECOVERY FROM ACUTE RENAL FAILURE AFTER 23 DAYS OFANURIA. DOUGHERTY JC, ET AL. JAMA 186:68 -9,_5 OCT 63HIGH CARBOHYDRATE INTAKE FOR ORAL USE IN ACUTE RENALFAILURE. PARSONS FM, ET AL. LANCET 2:386.8, 24 AUG 63ACUTE RENAL FAILURE. MUIR RD.PROC MINE MED OFFICERS ASS 42:89.95, JAN -APR 63HEMOLYTIC TRANSFUSION REACTIONS AND ACUTE RENALFAILURE: BLUEMLE LW JR.TRANS AMER CLIN CLIMAT ASS 74:201.10. 1962(VASCULAR DAMAGE IN THE BRAIN IN UREMIA) SZEGEDY L.PSYCHIAT NEUROL (BASEL( 146:116 -27, 1963 (GER)(RAPID LABORATORY DIAGNOSIS OF RENAL INSUFFICIENCY TNCLINICAL. PRACTICE AND IN THE HOSPITAL) BONITZ K. ET AL.THER MUIR 102:810-5, JUL 63' (GER)
ADAPTATION, OCULAR (G1)
NEURAL AND PHOTOCHEMICAL MECHANISMS OF VISUALADAPTATION IN THE RAT. DOWLING JE.J GEN PHYSIOL 46:1287-301, JUL 63
ADAPTATION, PHYSIOLOGICAL (G11
THE EFFECTS OF PREANESTHETIC MEDICATION, ANESTHESIAAND HYPOTHERMIA ON THE ENDOCRINE RESPONSE TO INJURY.VAN BRUNT EE, ET AL.ANESTHESIOLOGY 24:500-14, JUL -AUG 63 (7 REF.)
ON THE MECHANISM OF SOME TEMPERATURE EFFECTS ONDROSOPHILA. MILKMAN R.J GEN PHYSIOL 46:1151-70, JUL 63(STUDIES ON CHANGE IN THE ENERGY STEADY STATE INPHYSIOLOGICAL ADAPTATION PROCESSES. II. PRINCIPLES FORQUANTITATIVE EVALUATION OF THE PROCESSES IN
ADENOCARCINOMA
TEMPERATURE ADAPTATION USING AN ELECTRICAL ANALOGCOMPUTER METHOD) ZERBST E.PFLUEGER ARCH GES PHYSIOL 277:446.57, 9 AUG 63 (GER)(STUDIES ON CHANGES IN THE ENERGY STEADY STATE INPHYSIOLOGICAL ADAPTATION PROCESSES. I. DEDUCTION OF ANELECTRICAL ANALOG COMPUTER METHOD FOR ANALYSIS OF THEPROCESS OF TEMPERATURE ADAPTATION) ZERBST E.PFLUEGER ARCH GES PHYSIOL 277:434.45, 9 AUG 63 (GER)
ADAPTATION, PSYCHOLOGICAL(F, G1)
SELF - ESTEEM AND ADAPTATION. GRINKER RR JR.ARCH GEN PSYCHIAT (CHICAGO) 9:414 -8. OCT 63EMOTIONAL SYMPTOMS IN EXTREMELY ISOLATED GROUPS.GUNDERSON EK.ARCH GEN PSYCHIAT (CHICAGO) 9:362.8. OCT 63SEX DIFFERENCES IN RESPONSE TO SHORT.TERM SENSORYDEPRIVATION AND ISOLATION. ARHHOFF FN, ET AL.PERCEPT MOTOR SKILLS 17:81.2, AUG 63
NAMATJIRA: FAMOUS AUSTRALIAN ARTIST. PORTEUS SD.PERCEPT MOTOR SKILLS 17:13.4, AUG 63
ORDER, SEQUENCE, AND RETEST EFFECTS WITH THE S.1-0PROCEDURE. WILLIAMS JE.PERCEPT MOTOR SKILLS 17:3 -11, AUG 63
ADDISON'S DISEASE (C7)
THE POLYURIC SYNDROMES. BERMAN LB.GP 28:103.7, SEP 63
ADENINE (D2)
FREE NUCLEOTIDES OF THE NERVOUS TISSUE. 11. 'REENUCLEOTIDES IN THE BRAIN AND LIVER OF GUI' A PIGS.PECHAN I. ET AL. BIOLOGIA (BRATISL) 18:371.84. 1963IDENTIFICATION OF STIMULATORY FACTOR INVOLVED IHSYMBIOTIC GROWTH OF STREPTOCOCCUS LACTIS AlDSTREPTOCOCCUS CREMORIS. DAHIYA RS, ET AL.J BACT 85:585.9, MAR 63IDENTIFICATION OF GROWTH STIMULANTS FOR ST_CPTOCOCCUSLACTIS. KOBURGER JA, ET AL. J BACT 85:1051.5, MAY 63
CONFIGURATION AND BASE COMPOSITION OFDEOXYRIBONUCLEIC ACID FROM SPORES OF BACILLUS SUBTILISVAR. NIGER. MANDEL M. ET AL. J BACT 85:1445.6. JUN 63
BASE COMPOSITION OF THE DEOXYRIBONUCLEIC ACID OFSULFATE.REDUCING BACTERIA. SIGAL N. ET A'J BACT 85:1315.8, JUN 63
ADENINE NUCLEOTIDES (D5. D61
ROLE OF ATP -ADP EXCHANGE REACTION IN OXIDATIVEPHOSPHORYLATION. WADKINS CL, ET AL.FED PROC 22:1092.6, JUL -AUG 63NUTRITION AND METABOLISM OF MARINE BACTERIA. XII. IONACTIVATION OF ADENOSINE TRIPHOSPHATASE IN MEMBRANESOF MARINE BACTERIAL CELLS. DRAPEAU GR, ET AL.J BACT 85:1413.9, JUN 63
CHROMATOGRAPHY ON COMPOUNDS OF BIOLOGICAL INTEREST OFGLASS FIBER, PARAFFIN.COATED, AND UNTREATED CELLULOSEPAPER. EDWARDS CH, ET AL.J CHROMATOGR 11:349.54, JUL 63(REACTION RATE OF NADP - DEPENDENT OXIDOREDUCTASES INRAT BRAIN WITH3-ACETYLPYRIDINE.ADENINE.DINUCLEOTIDEPHOSPHATE)COPER H. ET AL. J NEUROCHEM 10:513.22, JUL 63 (GER)(CLINICAL AND EXPERIMENTAL GALACTOSEMIA. BIOCHEMICALASPECTS) KORC I.ARCH PEDIAT URUG 34:397.408 CONTD, JUL 63 (SP)
UENOCARCINOMA (C2)
OBSTRUCTIhG CARCINOMA WITH ACUTE PROXIMAL ULCERATIVECOLITIS. ADDLEMAN W.AMER J 'GASTROENT 40:174.8. AUG 63
PRIMARY MALIGNANT NEOPLASMS OF THE DUODENUM: REPORTOF TWELVE CASES. SPINAllOLA AJ, ET AL.AMER SURG 29:405.12, JUN 63VASCULAR TREE OF INTRAOCULAR TRANSPLANTS OF FROG RENALADENOCARCINOMA. WARNER L.ANGIOLOGT 14:116.20, MAR 63
PERCUTANEOUS TRANSHEPTIC CHOLANGIOGRAPHY AS ADIAGNOSTIC AID IN DISEASES OF THE BILIARY TRACT. MUMSA MI. ANN CHIR GYNAEC FENN 521SUPPL121:1.18/ 1963SPONTANEOUS REGRESSION OF PULMONARY METASTASESSECONDARY TO CARCINOMA OF KIDNEY. SAKULA A.BRIT J DIS CHEST 57:147.52, JUL 63
POSTMASTCCTOMY LYMPHOSTATIC ENDOTHELIOMA OFSTEWART AND TREVES IN A MALE. OETTLE AG, ET AL.BRIT J SURG 50:736.43, JUL 63
S5
ADDITIONAL INFORMATION ON SUBJECT HEADINGS, INCLUDING C1t3SS REFERENCES, WILL BE FOUND IN PART II OF THE JANUARY ISSUE.THE NUMBERS IN PARENTHESES FOLLOWING THE SUBJECT HEADINGS REFER TO THEIR LOCATION IN THE CATEGORIZED LISTING.
FIGURE 26. -Specimen Page of Jan..1964 Index Medicus, Produced on Honeywell Computer Printer.
52
Index Medicus
-(Herediatry diseases of lipid metabolism) Cohen AL.Se Med (B Air) 123:1576-85, 14 Nov 63 (Sp)
-(A case of xanthomatosis in a 4 1/2 year-old child)Shteinberg TA. Pediat Akush Ginek 6:33-4, 1963 (Uk)
XANTHOPHYLL (D6)
- A rapid method for the identification of smallquantities of lipid-soluble vitamins and quinones inbiological material. Lichtenthaler HK.J Chromatogr 13:166-72, Jan 64
-(The action of thioctic acid and helenien on visualacuity in reduced iSghting in normal subjects) CilottiP. Ann Ottal 89:1000-4v Dec 63 (It)
XENON (D1)
- Effects of xenon at elevated pressures in the dog.Domino EF, et al. Anesthesiology 25:43-53, Jan-Feb 64
- Interactions of lipids and proteins with anestheticgases. Huehlbaecher C, et al.Int Anesth Clin 1:937-52, Aug 63
-The provocative role of xenon anesthesia. Pittinger CB.Int Anesth Clin 1:963-8, Aug 63
-General biological significance of metabolically inertgases. Schreiner HR. Int Anesth Clin 1:919-26, Aug 63
- Measurement of blood-flow through skeletal muscle byintramuscular injection of xenon-133. Lassen NA, etal. Lancet 1:686-9, 28 Mar 64
- Modification of radiosensitivity of mice by inert gasesand nitrous oxide. Evans JC, et al.Radiat Res 21:243-55, Feb 64
XEROPHTHALMIA (C11)
- (Bitot s spots)Med Periskop 13:165-6, Dec 63 (14 ref.) (Ger)
XYLENE (D2)
- Gas-solid adsorption chromatography of some aromatichydrocarbons and nitrogen heterocycles on alumina.Klemm LH, et al. J Chromatogr 13:40-7, Jan 64
XYLOSE (D6)
- Specificity of sugar transport by the small intestine ofthe bullfrog. Rana catesbeiana. Lawrence AL.Comp Biochem Physiol 9:69-7J, May 63
-Insulin action and protein synthesis in diaphragmmuscle. Carlin H, et al.Proc Soc Exp Biol Med 115:127-9, Jan 64
-(The fat xylose test as a combined function diagnosis ofpancreatic and small intestinal diseases) BosseckertH, et al.Deutsch Z Verdau Stoffwechselkr 23:127-37, Oct 63
(Ger)-(The fat xylose test as a combined function diagnosis of
pancreatic ans small intestinal diseases) BosseckertH, et al.Deutsch Z Verdau Stoffwechselkr 23:127-37, Oct 63
(Ger)
YAWS (C1)
-Skin disease in school children in St. Lucia. Lees RE, etal. W Indian Med J 12:265-7, Dec 63
YEAST, DRIED (D5)
-Hypertensive crisis associated withnonoamine-oxidase inhibitors. Blackwell B, et al.Lancet 1:722-3, 28 Mar 64
YEASTS (135)
-New melibiose-utilizing yeasts, isolated fromalpechin . Santa Maria J.
Antonie Leeuwenhoek 29:329-43, 1963-The activation of yeast
uridinediphosphogalactose-4-epimerase by spermineand other cations. Darrow RA, et al.J Biol Chem.239:362-3, Jan 64
- Species specificity of amino acid acceptor ribonucleicacid and aminoacyl soluble ribonucleic acidsynthetases. Doctor BP, et al.J Biol Chem 238:3677-81, Nov 63
-Physiological studies in the black yeasts. Cooke WB, etal. Mycopathologia 21:225-71, 30 Dec 63
-Electron microscopical studies of frozen-dried yeast.
ZINC
IV. Schizosaccharomyces, Nadsonia andSaccharomycodes. Mundkur B.Z Naturforsch (B) 18:1073-82, Dec 63
- (On the occurrence of yeasts in the gastrointestinaltract with special reference to qualitativedifferences) Ott C. Mykosen 6:7-11, 1 Apr 63 (Ger)
-(The effect of gamma radiation (Co 60) and vitamin K5on yeast cultures) Gaisch H, et al.Z Naturforsch (B) 18:1070-2, Dec 63 (Ger)
-(The effect of sodium nitrite on nucleic acid bases andIts relation to the inactivation of yeast cells. 1. Thedeaminatien of nucleic acid bases in vitro and in vivo)Lochmann ER, et al.Z NaturlOrsch (B) 18:809-16, Oct 63 (Ger)
-(On the creetin-creatinine assimilation in yeast--fungusdiagnosis) Staib F.Zbl Bakt (Orig) 191:429-32, Dec 63 (Ger)
YELLOW FEVER (C1)
- Black men and malignant fevers. Berry LN.J Nat Med Ass 56:43-7, Jan 64
-Yellow fever. Burke-Gaffney HJ.Trop Dis Bull 61:113-8, Feb 64
YOHIMBINE (D4)
-(nesearch on the action of yohimbine in the guinea pig)Hazard R, et al.C R Soc Biol (Paris) 157:1915-7, 1963 (Fr)
-(Impotentia coeundi as an injury due to chemicalwarfare agents) Schirren C.Z Haut Geschlechtskr 34:189-95, 1 Apr 63 (Ger)
- (Cortical responses to the stimulation of the caudatenucleus. II. Modifications of the response to a singleimpulse induced by the administration of substanceswith synchronizing and desynchronizing action)Tartara A, et al.Boll Soc Ital Biol Sper 39:1137-40, 15 Oct 63 (It)
YTTRIUM (D1)
-Sr90 + Y90 bremesstrahlung efficiency in bone. LloydRD, et al. Int J Radiat Biol 7:339-51, Oct 63
YTTRIUM ISOTOPES (D1)
-Autologous bone marrow transplantation studies indogs irradiated by Y90 -DTPA urine-recycling technic.Winchell HS, et al. Blood 23:44-52, Jan 64
- The new chelating agent Ca-DTPA in the treatment ofprimary haemochromatosis. Kemble JV.Guy Hosp Rep 113:68-73, .1964
-A method for maintenance of Curie quantities ofY90 -DTPA in tl'e human body for defined timeperiods: Technique and dosimetry. Winchell HS, et al.J Nucl Med 5:16-26, Jan 64
-Radioactive isotopes in medicine. Therapeutic uses ofradioactive isotopes. Dickson RJ.London Clin Med J 5:67-77 concl, Jan 6?
- (Hypophyseal tumors after bilateral adrenalectomy inhyperfunction Cushing s syndrome) Bayer JM, et al.Deutsch Med Wschr 89:464-74, 6 Mar 64 (Ger)
- (Treatment with radioisotopes) Maeda T.Fukuoka Acta Med 54:1167-79, Nov 63 (Jap)
ZINC (D1)
- Zinc deficiency in man. Halsted JA, et al.Israel Med J 22:307-15, Sep-Oct 63
- Zinc deficiency syndrome in the young lamb. Ott EA, etal. J Nutr 82:41-50, Jan 64
- Analysis of August rat liver for calcium, copper, iron,magnesium, mangancrT, molybdenum, potassium,sodium and zinc. Everett JL, et al.J Pharm Pharmacol 16:85-90, Feb'64
- Treatment of multiple cutaneous carcinomas of theface by Mohs s chemosurgery method. Phelan JT, etal. New York J Med 64:410-3, 1 Feb 64
- The application of laser spectroscopy for thequalitative and quantitative analyses of the inorganiccomponents of calcified tissues. Goldman HM, et al.Oral Surg 17:102-3, Jan 64
- Metabolic interrelationships in the utilization of traceelements. Mills CF.Proc Nutr Soc 23 :38 -45, 1964 (53 ref.)
-(Our experiences with the zinc sulfate reaction incerebrospinal fluid) Vymazal J.Cesk Neurol 26:394-8, Nov 63 (Cz)
- (Simultaneous determination of cobalt, zinc, nickel and
S-4331
ADDITIONAL INFORMATION ON SUBJECT HEADINGS, INCLUDING (ROSS REFERENCES, WILL BE FOUND IN TART 11 OF THE JANUARY ISSUE.THE NUMBERS IN PARENTHESES FOLLOWING THE SUBJECT HEADINGS REFER TO THEIR LOCATION IN THE CATEGORIZED LISTING.
FIGURE 27.-Specimen Page of July 1964 Index Medicus, Produced on IBM 1403 Printer.
53
face a major decision: delay the inaugurationof MEDLARS or find an alternate way toprint Index Medicus.
A decision was made to follow the originalproduction schedule, and a crash program-ming effort resulted in a set of programs toprint Index Medicus on the standard com-puter printer with a very limited characterset of uppercase symbols only, using thesame tapes (EPCF) for input that wouldhave been used to produce GRACE copy. Thisprocedure was used for six issues, Januaryto June 1964 (see Figure 26).
During the first part of 1964, these printprograms were modified to create a tape forprinting on an IBM-1403 Chain Printer at aprivate company in Washington, D.C. Thisprinter was capable of producing both upper-and lowercase letters and the July 1964issue was produced by this method (see Fig-ure 27).
One year late, GRACE finally arrived atNLM in May 1964 and, after careful check-out, was used for production of Index Medi-cos beginning with the August 1964 issue.The decision to begin production in January1964 proved to be a good one, since theEPCF's for the first seven issues producedby the computer printer were available foruse in producing the Cumulated Index Medi-cus for 1964 by GRACE. Note: Refer back toFigure 21 to compare GRACE page withHoneywell and IBM printout.
Initial Production PeriodAs described above, MEDLARS became op-
erational in January 1964. Transition to thenew system was effected without great diffi-culty, a testimony to the joint efforts of thecontractor and NLM staff in system testingand training of personnel. Contractor person-nel (except those working on GRACE) wereable to leave the Library in February 1964and no system maintenance contracts wererequired.
Computer programs of the Input and Pub-lication Subsystems seemed to work muchmore effectively during the initial year ofproduction than did the Demand Search Pro-grams. Many program "bugs" continued toappear in the search routines. These prob-lems, coupled with problems in the conver-
54
sion CCF file described earlier, resulted inpoor demand search production in 1964.
Some of the major deficiencies in theMeSH vocabulary were discovered in 1964.The Library began to realize that outsideexpert advice was needed for improving thevocabulary, and major revisions were madefor 1965. Other problems were encounteredin smoothing out operating procedures; suchthings as error and rejection rates of inputcitations were found to have been underesti-mated.
Maintenance of the System;Changes Made
No system design is ever perfect; and nogood manager will be content to have a sys-tem remain static without constant improve-ments and changes being made. MEDLARSis no exception to this rule. Many problemsencountered during operation have resultedin changes from the original design. Some ofthe more important changes are listed below:
(1) Input Errors. The high number of er-rors in citations entering the systempointed up the need for additional in-put editing. As a result, a new proce-dure was installed to print all newcitations added to the CCF each dayand have these proofread by an expe-rienced editor.
(2) Change of Recurring BibliographyProcedures. The original input pro-grams created both a CCF and PCF(Processed Citation File). Selectionand formatting of citations for recur-ring bibliographies was done by theInput Module, and the PCF was useddirectly in the Publication Subsystem.This approach was believed to be effi-cient in that it leveled out the outputworkload by doing it every day. How-ever, in actual practice, this methodturned out to be very inefficient be-cause of the long Input Module runsthat resulted. In 1966 the Library con-verted to a single file system (CCF)and used the Demand Search Pro-grams to select citations for recurringbibliographies. The result was a netsaving of over fifty hours per monthof computer time.
50
40
30
20
10
AVERAGE PROCESSING TIME PER SEARCH
Minutes
APR-JUN FY'67 SEP DEC MAR JUN
FIGURE 28.
(3) Demand Search Problems. High rejec-tion rates for demand searches alongwith slow turnaround times causedmany changes to be made in the Re-trieval Subsystem. Conversion fromcomplete paper tape punching of allsearch formulations to use of pre-Punched MeSH cards (see Chapter IV)improved the error and rejection prob-lem. Major changes in the DemandSearch Programs resulted in an in-crease of average batch size fromabout ten searches per batch to asmany as 60, with the average between25 and 30. The computer processingtime for a search was reduced toabout 30 % of its previous length. Fig-ure 28 is a chart showing the improve-
FY'68 SEP DEC MAR
17
JUN
ments in search processing time overthe years. Procedural changes havealso helped to improve turnaroundtime somewhat.
(4) File Maintenance. Many inefficiencieswere discovered in the File Mainte-nance Module. Perhaps the most se-rious was the inability of these pro-grams to handle mass changes in thetree structures (hierarchy of terms)of MeSH. Modification of the moduleto add this feature resulted in a 9-to-1improvement in computer time, and,perhaps more important, a reductionof months of clerical effort previouslyrequired to change individual ci-
tations.
55
HUNDREDS
25--
20
15
10 -
5
0
(5)
DEMAND SEARCHES COMPLETED-1st QUARTER, FY 1968
RECEIVED BY
1000NLM
500
MEDLARS
STATIONS
'100.'NIH'
..I60,:01-110:., ..,;x:1;
,SODEN,
tn"-',-,---iv-s:,,.*4.1-'' 4'4.1.,^: 1
M, :Pi::
-1-;:;.-'7*%
,!-.At,.,
FORMULATED &
REVIEWED BY
600NLM
900.
MEDLARS
STATIONS
100.NI
,..,,....., .h.10,....g.
FIGURE 29.
MeSH Revision. The original systemdesign for MEDLARS grossly under-estimated the size and number ofchanges to MeSH that would be re-quired for the annual revision. Newprocedures had to be developed, withutility programs written to aid thestaff of the MeSH Section. Over 500new terms have been added for the1968 revision.
Other changes are still required and systemmaintenance will be an on-going process. TheNLM 'programming staff must constantlyplan on system maintenance as a major fac-tor in estimating workloads.
MEDLARS DecentralizationDecentralization of the MEDLARS Re-
trieval Subsystem was planned during theoriginal systems design period. The major ob-jectives of the decentralization program
COMPUTER
WORK BY
were : (1) improve demand search serviceby providing regional service closer to usersthroughout the country; and (2) increase thetotal number of searches processed each yearby the use of multiple search stations.
Consideration was given to several alter-nate approaches to decentralization, includ-ing: (1) decentralize the search formulationactivities, but continue centralized computerprocessing at NLM, and submit searches viadata communication equipment; (2) estab-lish regional offices of NLM with both search-ers and their own computer equipment forsearch processing; and (3) contract withexisting centers to provide search serviceson a regional basis. After much study, thelast was selected, since it would bring theservice closer to the user, and, at the sametime, bolster local medical library resources.The original plans thus developed called forcontracts to be awarded to from six to ten
universities or other nonprofit organizationswith a combination of good medical libraryand computer resources.
Two contracts were awarded in 1965one with the University of Colorado toserve the Mountain Time Zone, and anotherwith the University of California at Los An-geles to serve the Far West. The ColoradoMEDLARS staff contracted for time on aHoneywell-800 computer at the Denver Fed-eral Center, and, since no major reprogram-ming was needed, was able to become opera-tional in short order. The Colorado Stationhas been .running searches since February1965.
The UCLA contract was more complex.Computer processing was to be done on anIBM 7094-7040 direct coupled computersystem. Since complete reprogramming ofthe Demand Search and Report GeneratorModules was required, the Library directedthat these programs be rewritten in COBOL,a machine-independent language which couldbe used on other makes and models of com-puters. This decision turned out to be a mis-take, at least in part. Since the COBOL lan-guage is necessarily generalized, it does notgenerate the most efficient operating pro-grams. The combined high-speed logicalsearch program took so much computer timethat it had to be rewritten in a machine-oriented language. Unfortunately, the MED-LARS computer system at UCLA has neverbecome operational. Delays encountered inreprogramming and the replacement of the7094 computer by an IBM-360/75 beforethe project was completed caused NLM staffto change plans and process UCLA MED-LARS searches on the central computer inBethesda.
Contracts were awarded in 1966 to Har-vard University, the University of Alabama,and the University of Michigan for develop-ment of additional MEDLARS Stations,bringing the total to five. The Stations havesent personnel to NLM for six months of in-tensive training in indexing, MeSH, andsearch formulation. All five Stations are nowformulating searches. However, because ofreprogramming problems, only the ColoradoStation was actually processing searches asof January 1, 1968 the other Stations mail-
ing formulated searches to Bethesda. Someof the reasons for this are discussed below.
In addition to tit.: five regional MEDLARSStations, Ohio State University and the TexasMedical Center in Houston asked for accessto the MEDLARS tapes in order to establishstations with their own resources. Ohio State,which has two IBM-7094 systems, completeddebugging of the UCLA MEDLARS Pro-gram, made some modifications, and now hasthe system operational. Ohio State was re--,ently designated as MEDLARS Search Sta-tion for the State of Ohio, and the TexasMedical Center is now operating privatelywithout financial support from NLM.
MEDLARS decentralization has also takenplace internationally. The United Kingdom isnow operating a MEDLARS Station for allof Great Britain. Searches are formulatedat the National Lending Library for Scienceand Technology, and computer processing ishandled at the University of Newcastle-upon-Tyne. Complete reprogramming of thesearch routines was successfully accom-plished for the English Electric KDF-9computer.* A second international MED-LARS Station is in successful operation atthe Karolinska Institut in Stockholm,Sweden. Searches at this Station are pro-cessed on a system utilizing an IBM-1401and IBM-7094 computer.
Figure 29 is a graph showing searches pro-duced during one three-month period throughthe MEDLARS Decentralization Program.Plans have also been made to include MED-LARS Search Stations at all Regional Medi-cal Libraries to ;32 established under the pro-visions of tha. Medical Library AssistanceAct of 1965.
It is obvious from the above discus-sion that the MEDLARS DecentralizationProgram has had many problems. Some ofthese are summarized below:
(1) Reprogramming Problems. The timeand effort involved was badly under-estimated by NLM staff.
(2) Tape Conversion Problems. The prob-
*A. J. Harley and E. D. Barraclough, MEDLARSInformation Retrieval in Britain, The PostgraduateMedical Journal, London, Feb. 1966.
57
lem of converting from one magnetictape form at to another was thoughtto be trivial, but this was not so.Many problems were encountered; forexample, a missiag designator indicat-ing the end of a data field on a tapegenerated by NLM causes havoc withthe Tape Conversion Program at aMEDLARS Station.
(3) Difficulties Encountered at UniversityComputer Stations. Most universitycenters have not had experience withthe operation of large -file data pro-cessing systems. They are more accus-tomed to running short computationalproblems. One Station didn't evenrealize that it had a defective mag-netic tape drive which was stretch-ing tapes until it started MEDLARSprocessing, Similar problems wereencountered in programming andsoftware development.
(4) Communications Problems. Because ofthe remote location of the MEDLARStations, communications betweenNLM and regional personnel were notalways as good as they should havebeen. The coordination and training ofsearchers turned out quite well, how-ever.Because of the above, developmentaltime and costs for the Stations werehigher than originally anticipated.
(5)
Formal Evaluation of MEDLARSA comprehensive project to evaluate the
demand search performance of MEDLARSwas initiated in January 1966two yearsafter operation began. A preliminary testwas conducted early that year to establishthe feasibility of approach, and the full testprogram became operational in August 1966.The evaluation was completed in December1967, and the final report will be publishedin 1968. An advisory committee of six expertsin the fields of documentation, statistics, in-formation retrieval, and computer scienceprovided a critical review of the study as itprogressed.*
Twenty organizations, representing theprincipal types of medical groups (academic,
58
atnical, research, pharmaceutical, regula-tory) making use of MEDLARS, cooperatedin the study. The majority of the demandsearch requests made by members of theseorganizations within a 12-month period weretreated as test requests. These were all "real-life" requests, representing actual informa-tion 'needs. While the organizations agreedin advance to cooperate in the study, theindividual requester knew nothing of theevaluation until his request had been madeto the system.
The test program was conducted to deter-mine how well MEDLARS retrievals aresatisfying the demands of users and, in addi-tion, what factors are the principal contribu-tors to system failures. Data derived fromthe test program are of two types : (1) per-formance figures and (2) case studies.
Performance figures include recall ratios(the proportion of the total of relevant arti-cles, contained in the data base, that is re-trieved in a search), precision ratios (theproportion of all articles returned in a searchthat are judged relevant), and novelty ratios(the proportion of the relevant articles re-trieved that are brought to the requester'sattention for the first time by the MEDLARSsearch). Recall and precision ratios are usedto compare the performance of the systemin various modes and conditions of operation.For example, they allow a comparison of per-formance for various kinds of user groups,for various kinds of requests (across broadsubject fields), and for requests handled withvarying levels of user-system interaction.
Case studies are analyses of system fail-ures (i.e., failures to retrieve known rele-vant articles, or instances in which largenumbers of irrelevant citations were re-trieved). An analysis is conducted to deter-mine at what point the system failed.
*F. W. Lancaster, Evaluating the Performanceof a Large Operating Retrieval System, Proceed-ings of the Sccond Electronic information HandlingConference, Thompson Book Co., Washington, D.C.,1967.
The evaluation program is diagnostic, in-tended to identify major problem areas, andthus allows appropriate corrective action insystem design and operating procedures.
The study revealed that MEDLARS is op-erating, on the average, at about 58% recalland 50% precision. Thus, the system re-trieves about 58% of all the references inthe system which would be of value to therequester, and one out of every two docu-ments retrieved is judged to be of value bythe requester. The study further showed con-siderable scatter of results, indicating widevariation of performance among the testsearches.
Of considerably more importance thanthese figures are the reasons for system fail-ures. Each system failure was extensivelyanalyzed and attributed to the indexing, in-dexing language, user-system interaction,searching strategies employed, computerprocessing, or a combination of the above.The results of this analysis, based on theresults from 300 test searches, are listed inTable 7. Further discussion of this analysiscan be found in the final report of theMEDLARS evaluation project.*
Table 7Failure Analysis Table
Number of Nu, tber ofFailure Attributed To: Recall Precision
Failures Failures
Indexing
Indexing Language
298 393
81 1094
Inadequate User-SystemInteraction 199 503
Searching Strategies 279 983
Computer Processing 11 3
Performance in Relation toOriginal System Objectives
To summarize this chapter on operatingexperience, the original system objectives ofMEDLARS (as listed in Chapter I) are re-examined to see how well they have beenmet and what changes in goals were re-quired:
Objective No. 1:Improve the quality of and enlarge (broad-en the scope of) Index Medicus and at thesame time reduce the time required to pre-pare the monthly editions for printingfrom 22 to 5 workdays.
Comment: For the most part, this objec-tive has been met. The quality of IndexMedicus has been improved through im-provements to MeSH, improvements in thejournal selection process, and constant at-tention to indexing accuracy. The scope hasbeen broadened considerably from 129,808articles from 2,275 journal titles in 1963to 163,077 articles from 2300 journal ti-tles in 1967. Processing time has been cutto about five days.
Objective No. 2:Make possible the production of other com-pilalions similar to Index Medicus in formand content (but in more specific medicalsubject areas and hence smaller in size).
Comment: This objective has been accom-plished through the production of recurringbibliographies (see Table 1, Chapter II).However, the original estimate of 50 re-curring bibliographies was too high andonly nine were in production on January 1,1968, with some others in the planningstage. Because of the difficulties in theintellectual effort required, the develop-ment of recurring bibliographies turnedout to be a much larger job than wasoriginally estimated.
Objective No. 3:Make it possible to include in Index Medi-cus and other compilations citations de-rived, not only from. journal articles, butfrom other sources as well.
Comment: The system capability to in-clude citations to monographs in MED-LARS was provided through reprogram-ming for the Current Catalog (see Chap-ter VIII).
*F. W. Lancaster, Evaluation of the MEDLARSDemand Search Service, National Library of Medi-cine, Bethesda, Maryland, 1968.
59
Objective No. 4:Make possible the prompt (a maximum oftwo days) and efficient servicing of re-quests for special bibliographies, on botha demand and a recurring basis, regularlysearching up' to five years of stored com-puter files.
Comment : It has not been possible to ac-complish a two-day search turnaroundtime with the present batch processingsystem. Also, because of computer time limi-tation, demand searches beginning in early1968 include only references from January1966 on, instead of references from theentire file.Objective No. 5:Increase the average depth of indexing perarticle (number of descriptive subjectterms per article) by a factor of five, i.e.,ten headings versus two.
Comment: This objective has been accom-plished for depth journals to which an av-erage of about ten subject headings perarticle are assigned (see Chapter III) .However, non-depth journals are indexedto an average depth of only about fourterms.
Objective No. 6:Nearly double the number of articles thatmay be handled (indexed and entered intothe computer) annuallyfrom 140,000now to 250,000 in 1969.Comment: Production has increased withMEDLARS to 165,000 articles input in1967. Library management hopes to meetthe goal of 250,000 articles by 1969, butthe attainment of this goal remains un-certain.Objective No. 7:Reduce the need for duplicative total lit-erature screening operations (at other li-braries and information centers).
60
Comment: This objective is difficult to eval-uate. However, all indications point to itsbeing met. index Medicus and other MED-LARS bibliographies are used extensivelyin libraries and information centers as lit-erature screening tools. No other medicalindexing operation comparable in size toMEDLARS has come into existence. Hence,it can reasonably be assumed that a gond
deal od, :,xpensive, duplicative indexing hasbeen avoided.
Objective No. 8:Keep statistics and perform analyses ofits own operations to provide the informa-tion needed to monitor and improve sys-tem effectiveness.
Comment : This objective has been par-tially accomplished through the StatisticalModule Reports described in Chapter VI.Other reports have been identified thatcould be derived from the system andcould be very useful (e.g., statistical dataon demand searches). However, for themost part, NLM programmers have beentoo busy to work on these reports.
Objective No. 9:Permit future expansion to incorporatenew and as yet not completely definedhence, secondaryobjectives.
Comment: This objective has been accom-plished. Installation of MEDLARS hasprovided a base of data processing equip-ment and experienced system personnel atthe Library for work on new systems andextensions of MEDLARS. Chapter VIIIdescribes a major new system, derivedfrom MEDLARS, which has been in pro-duction since January 1966.
Of the nine original objectives of MED-LARS, three have been completely accom-plished, six partly accomplished.
Chapter VIII
THE CURRENT CATALOGAN EXTENSION OF MEDLARS
Background and History *The original plan for MEDLARS called
for entry of cataloging data for all newmonographs, theses, and serial titles into thecomputer files. Citations representing thiscataloging were to appear in Index Medicus,which would constitute the only book-formrecord of the NLM catalog. This plan wassubsequently abandoned for several reasons.First, systems limitations in MEDLARSwould not permit the publishing of citationsin acceptable form without extensive modi-fications. Second, the books and serial titlesare not assigned subject headings in suffi-cient depth to be effectively mixed with themore detailed indexing of journal articles.Third, a re-examination of the potential usesof a book-form catalog pointed up possibleadvantages to other medical libraries in aseparate book-form catalog printed at fre-quent intervals.
For these reasons, the original plan wasdropped, and, in February 1964, a systemsanalyst began work on a new computer-aidedcatalog processing system. This system rep-resents the first major extension of MED-LARS, taking further advantage of the ex-perienced data processing staff and equip-ment acquired to operate MEDLARS. Nooutside contract help was used in the designor implementation of the cataloging system.
Design work lasted for about one year andprogramming began early in 1965. Ratherthan writing all new programs, the new sys-tem relied heavily on modifications to the
*For a more detailed description of this systemsee: I. J. Weiss and E. V. Wiggins, Computer-Aided Centralized Cataloging at the National Li-brary of Medicine, Library Resources and TechnicalServices, Vol. 11, No. 1, Winter 1967.
MEDLARS Input and Output Modules to in-corporate new requirements. System testswere run during November and December of1965, and the new system went into pro-duction, on schedule, in January 1966.
The major objectives of this system are(1) to make NLM cataloging available toother medical libraries on a current and fre-quent basis for use as an acquisitions andcataloging tool, and (2) to improve the in-ternal production of cards for the centralNLM card catalog.
Products of the SystemThe cataloging system produces two major
products: the NLM Current Catalog andcatalog cards.
The NLM Current Catalog (see Figure 30)is computer-produced and published biweeklyand quarterly. The biweekly issues containcitations to publications cataloged by theLibrary which have recent imprint dates.The quarterly issues are cumulative and con-tain all citations cataloged from January ofthe current year, regardless of imprint date.The final quarterly cumulation of the yearis hardbound and serves as the permanentrecord of that year's cataloging.
Both biweekly and quarterly issues con-tain a Subject Section, with citations listedalphabetically by main entry under subjectheadings. Subheadings are used in the quar-terly issues only. All issues also contain aName Section, an alphabetic author/titlelisting of citations. Citations appear in fullonly under the main entry and cross-refer-ences appear only in the quarterly issues.Library of Congress card numbers, whenavailable, are included with the citation.
The second product of the system is thecatalog card. Complete 3" x 5" catalog card
61
NLM CURRENT CATALOG
THORACIC DISEASESUglov, Fedor Grigoevich. Oslozhnenlia prl
vnutrigrudrlykh operatsllakh. (Leningrad]Meditsina, 1966. 442 p. Illus. At head oftitle: F. G. Uglov (I dr.]WF 980 U2Go 1966 Cit. No. 41377
TOINIOGRAPHYFavez, G6rard. L'exploration radiologique du
poumon et du mddlastIn a l'alde de latomographle oblique postdricure a 55 (degrds]Radiological examination of the lung andmediastinum with the aid of posterior obliquetomography at an angle of 55 (degrees, par]Gdrard Favez (et] Omar Soliman. Basel, Kargcr.1966. 187 p. Illus. Text In French, English,and German.Wt' 141 F273e 1966 Cit. Nv. 41126
TOOTHDrum. Walter. Imagines demonstrandae.
Berlin. Verlag Die Qulntessen i19651 31 p., 58ff. e. 69] plates. Text and part of illustrativematter in pocket. Text in German.WU 17 D795I 1965 Cit. No. 41651
TOOTH DISEASES14bakov. Anatol!! lvanovich. ed. Spravochnik po
stomatologil, pod red. A. 1. Rybakova (I dr.)Moskva, Mcditsina, 1966. 318 p. Illus.WU 100-11989s 1966 Cit. No. 41400
TRACE ELEMENTSVsesoluznyl simpozium "Mikroelementy I nervnaia
sistema," Baku, 1963. Mated:11y._ (Red. A. 1.Karacv) Baku, Iztlvo Akademil nauk
zerbaftlzhanskof SSR, 1966. 131 p. Illus. Athead of title; Akademila nauk Soluza SSR.Vsesoluznoe fiziologicheskoe obshchestvo im. 1. P.Pavlova. Akademi la nauk AzerbafdzhanskotSSR. Sektor fizIologil.QU 130 V985m 1963 Cit. No. 36443
TRAINING SUPPORTU. S. National Institute of Mental Health. Public
Information Section. Training programs of theNational Institute of Mental Health. (Rev. ed.Bethesda, Md., 1966] 21 p. (Public HealthService publication, no. 1283)WM 18 U582t 1966 Cit. No. 42350
TRAVEL;.ggerbeek, Iver. Laegens reJser5d.
(Kobenhavn, Europaciske vate- ogreJsegodsforsikring, 19661 43 p. Illus.W6 P3 CIt. No. 41211
TROPICAL MEDICINEMeditsinskala geografila troptcneskikh stran. ((My.
red. sbornika, M. B. Gornung, A. M. Lebedev]Moskva, Myst*, 1965. 221 p. Hies. (Voprosygeografli, sb. 68) Some articles have summaryin English.WC 680 M491 1965 Cit. No. 41643
TUBERCULOSISBryzgalova, Olga Konstantinovna. Tuberkulez
Izlechim. Moskva, lzd-vo Meditsina. 1966. 22p. Illus. (Nauchno.popullarnala meditsinskalaItteratura)W6 P3 CIt. No. 41421
Dvotrin, :Mikhail Seincnovich. Tuberkulez v druantiblotikov; OpIdemiologlia I statlstika. Kiev,Zdorovia, 1966. 253 p.WF 200 D988t 1966 Cit. No. 41403
Nursing Advisory Service on it and Otherprwiratory Diseases. Exchange of ideasnationwide 1963965. New York, NationalLeague for Nursing, 1965. iv, 203 leaves.WY 163 N977e 1965 Cit. No. 41773
Todorov, Stefan. Tuberkuloza; prichlni zarazvltleto I, predpazvane I prolavi. Sof! la,Medltsina I fizkultura, 1966. 27 p. illus.(Bibiloteka "Zdravni besedi za scloto," 1966, 5)WF 200 T639t 1966 Cit. No. 42336
ULNAKnussmann, Rainer. Humerus. Ulna und Radius
der Simlac; vergielchend:norphologlscheUntersuchungen mit Beriicksichtigung derFunktiOn. Basel. New York, Karger. 19,37. lx,399 p. illus. (BIbilotbeca primatologica, lase. 5)WI 151428T fasc. 5 1967 Cit. No. 41352
UNIVERSITIESKilian, Hermann Friedrich. 18001863. Die
Universitacten Deutsch lands inmedicinischnaturwissenschaftlicher Ilinsicht.Amsterdam, Isradi. 1966. facsim.: v. 404 p.port. Enlarged reprint from the editionHeidelberg und Leipzig 1828.W 19 ii49u 1828F Cll. No. 41662
UROLOGIC DISEASESBeltSikov, Josep. Uroloogliised haigused; nende
.altimine Ja ray!. Tal lintt. Kirlastus Valgus.196G. 51 p. (Teadus Ja tervis)WG P3 Cit. No. 41533
VASCULAR DISEASESWhelan, Thomas J. Non atheroscierotie arterial
lesions and their management (by] Thomas J.Whelan (and] Joseph 11. Baugh. (Chicago. 1967]nos. illus. (Current problems in surgery,February, 1967. etc.)WI CU804S Feb. 1967 etc. Cit. No. 42351
VENA CAVA. SUPERIORHoward, Norman. Medlastinal obstruction in lung
cancer. Edinburgh, Livingstone, 1967. v11. 100p. illus.WF 500 11851m 1967 Cit. No. 41744
VETERINARY MEDICINE'Strong, Charles L. Horses' injuries;
common-sense therapy of muscles and Joints forthe layman. London, Faber and Faber (19671118 p. !HuiSF 901 S923h 1967 Cit. No. 41747
U. S. Agricultural Research Service. Healthrequirements and regulations governing theinterstate and international movement oflivestock and poultry, State (and] Federal.Revision Dee. 1965. (Hyattsville, Md., 1965] Iv. (looseJtaf) Its ARS 91.17-3)SF 623 U58h 1965 CIt. No. 42066
Veterinarnye instrumenty, apparaty, pribory Ioborudovanie; katalog. Moskva, V/OMedeksport (19667] 259 p. illus. Addedtitle pages in German, English, and French;contents In all four languages.SF 913 V586 1966 Cit. No. 41517
VIBRATIONAngelova, Manila. Vibrats!onna bolest. Sof! la,
Meditsina 1 fizkultura, 1956. 28 p. (Biblioteka"Zdravni besedl za rabotnika," 1906. 6)WA 100 A58v 1966 Cit. No, 42355
Malinskala, Nadezhda Nlkolaevna. Profilaktikavibratsionnykh zabolevanif. Moskva, Meditsina,1966. 35 p. Illus. (Nauchnopopullarnalamedit.sinskala literature) At head of title: N. N.Mallnskala, N. B. Met !Ina.W6 P3 CIt. No. 41427
VIRUSESKharizanova, Tanta. Na granitsata nn zhivota,
Solna, Narodna mladezh, 1966. 137 p. Illus.,ports. (131biloteka za politekhnicheski znanlia, no.29)QW 160 K44n 1966 Cit. No. 41633
Kukalne, Rita, ed. Virus I kletka. (Redkolleglia:R. A. Kukaln, otv. red., i dr.] Riga, Izti.voZlnatne, 1966. 308 p. Illus.QW 160 K98v 1966 ell. No. 41354
VITAMINSBersin, Theodor, 1902- Biochemie der
Vitamine. Frankfurt a. M., AkadernischeVerlagsgeselischaft, 1966. 296 p. Silas.QU 160 B535b 1966 Cit. No. 41506
Malinov, Todor. Kak da su nabaviamc vitamin!prez vsichki sezoni. Plovdiv, Danov, 1966. 121
p. (Poreditsa Zdravc)QU 160 M251k 1968 Cit. No. 41614
VIVISECTIONDreyfus. Marie. comp. Crimes againg creation.
Trowbridge, Eng.. Massey 11966] 322 Ji.IIV 4915 D778e 1966 Cit. No. 41137
VOCATIONAL GUIDANCEVaughan, Paul. Work to be done; careers in
mental health. London, National Association forMental Health (1966] 71 p. Illus.WM 21 V368w 1966 Cit. No. 41574
WATER POLLUTIONAmerican Fisheries Society. A symposium on
water quality criteria to protect aquatic life,presented at the 96th annual meeting, KansasCity, Missouri, September 1966. Edwin L. Cooper,editor. (LanTence. Kan.] 1967. vill, 37 p.illus. (its Special publication, no. 4)Supplement to vol. 96, no. 1 of Transactions ofthe American Fisheries Society.W6 P3 Cit. No. 41212
California. State Water Quality Control Board.An investigation on the fate of organic andinorganic wastes discharged into the marineenvironment and their effects on biologicalproductivity. (Sacramento] 1965. xi. 116 p.illus. (Its Publication. no. 29)WA 778 CI 31 1965 Cit. No. 37651
Colorado. Legislative Council. Water pollution inColorado; Legls:ltIve Council report to theColorado General Assembly. (Denver, 1965]xxix, 57 p. Illus. Om Research publication, no.105)WA 689 C719w 1965 Cit. No. 41210
WATER SUPPLYPan American Health Organization. Bombas para
agua potable; apuntes del curso intensivo.Washington, 1966. vii, 269 p. Illus. (PanAmerican Sanitary Bureau. Publicacionesclentiflos, no. 145)WI PA443 no.145 1966 Cit. No. 41657
Simposie v.ntw Agua Potable y Alcantarillado,Mexico. 1966. Memoria. (Nikko. Sc;:retariade Recursos lildraulicos, 1966) 454 p. Illus.,ports.WA 670 S609m 1966 Cit. No. 42133
WOMENTSaneva, Niko Una. Khiglena na zhcnata v
prolzvodstvoto. Sofila, Mcditsina I fizkultura,1966. 26_p. (Bibiloteka Zdravni besedi zarabotnika. 1966, 4)WA 491 T877k 1966 Cit. No. 41317
WORKLuklanov, Vladimir Sergeevich. Trud i zdorov'e.
Moskva, Znanie, 1967. 44 p. (Narodnyiuniversitet. Fakullet zdorovia, 1967, 1)W6 P3 CIt. No. 41647
WORKMEN'S COMPENSATIONBraga, Marigildo de Camargo. Moldstias
profisslopais c docncas do trabalho. Sit) Paulo,Edit Ora Aries (19669. v.\V 925 B813m Cit. No. 41326
Work Injury cases settled under the Workmen'scompensation act of Wisconsin: violations ofsafety rules and the Child labor law. 1965 -
Madison. -v. Continues injury usessettled under Workmen's compensation act ofWisconsin: violations of safety rules and theChild labor law. Issued by IndustrialCommission of Wisconsin, Statistical Division asits Statistical release, 3804, etc.W2 AW6 14wd Cit. No. 41750
WOUNDS AND INJURIESJamieson, Kenneth G. Traffic injury in Brisbane;
report of a general survey, by K. G. Jamieson and1. A. Tait. Canberra, 1966. ail, 353, ix p.Illus. (Australia. National Health and MedicalResearch Council. Special report series, no. 13)W2 KA8 N2s no.13 1966 Cit. No. 41375
62
FIGURE 30.-Specimen Page of Biweekly NLM Current Catalog.
1r
WT Emery, Ralph, 1920-120 Understanding old age. London, National Assn. forE53u Mental Health, 1967.1967 40 p. illus.
1. Geriatrics - popular works I. National Associationfor Mental Health, London. II. Title
National Libra
Understanding old age.
WT Emery, Ralph, 1920.120 Understanding old age. London, National Assn, orE53u Mental Health, 1967.1967 40 p. illus.
1. Geriatrics popular works I. National Associationfor Mental Health, London. II. Title
WT Emery, Ralph, 1920.120 Understanding old age. London, National Assn. forE53u Mental Health, 1967.1967 40 p. illus.
1. Geriatrics - popular works I. National Associationfor Mental Health, London. II. Title
National Libra
68.76066
National Association for Mental Health, London.
WT Emery, Ralph, 1920.120 Understanding old age. London, National Assn. forE53u Mental Health, 1967.1967 40 p. Illus.
1. Geriatrics - popular works I. National Associationfor Mental Health, London. II. Title
National Library of Medicine 68.76086
FIGURE 31.Specimen Computer-Produced Catalog Cards.
63
sets for each citation are produced by thecomputer for the central NLM fi)e (see Figure31). Cards are produced for the main entryand all added entries as well.
Except for the omission of price, informa-tion on the card is approximately the sameas that in the book catalog. The format isthat of a traditional catalog card, with appropriate indentations and call number inthe left corner.
Equipment UsedThe same basic equipment used for pro-
cessing journal article citations is also usedin the cataloging system. This includes theFleAowriter paper tape typewriters for in-put preparation, the Honeywell 800 and 200computers for processing, and GRACE foroutput printing of both the book catalog andcatalog cards. In addition, the system alsomakes use of a microfilm camera and Copy -flo reproducer in the production of catalogcards, as described below.
Information Flow and ProceduresFigures 32 and 32a are charts showing
the flow of information through the catalog-ing system.
The production of catalog cards and bookcatalogs is the joint responsibility of theTechnical Services Division and the Office ofCompater and Engineering Services. Thecatalogers' contribution is much the same asit has been; only now, the record of the tra-ditional information is in machine-readableform.
Entries for both the biweekly and quar-terly publications are in accordance withcurrent American Library Association rules,with a few exceptions : only the Roman al-phabet is used; titles in Cyrillic alphabetsare cited in transliterated form only; titlesin Oriental languages are translated as well.Average depth of subject cataloging is 1.5subject headings per item.
Input data sheets (see Figure 33), pre-pared by the catalogers, are collected oncea day, forwarded to the Office of Computerand Engineering Services, and typed in theform of hard copy and punched paper tape.The hard copies are visually proofread. Thecorrected hard copies are returned to the in-
64
put typists for preparation of correctiondata.
Original paper tapes and correction papertapes are spliced together. The correctionpaper tapes must always accompany theoriginal tapes in the same input run to beproperly processed by the computer.
The first six computer programs of theMEDLARS Input Module were modified toprocess cataloging data as well as journalarticle input data. The modified programs editthe citations and convert medical subjectheadings from English descriptors to theircomputer-coded equivalents.
Citations are checked within the computerfor certain types of errors. Fields (authors,titles, etc.) incompatible with the input pro-grams, with the exception of the subjectheadings field, cause the entire citation to berejected. If subject headings are incorrect,the update program accepts the citation withthe qualification that a "no-catalog card"flag is generated for that citation. Thismeans that no catalog cards will be gener-ated for the citation until file maintenance isperformed to correct the erroneous subjectheadings on the citation. All rejected cita-tions are listed with the reason for rejection.
All citations on the daily transaction tape,divided into main and subrecords for editingpurposes, are sorted and merged into a unitrecord. Each citation is assigned a sequentialnumber by the computer and is then mergedinto the cumulated catalog file (history);thus, the cumulated catalog file is in order bycitation number. As daily transactions aremerged into the history file, each citation ischecked for a "generate catalog card" flag.Those citations containing this flag arecopied on the card print fife. The Card Ex-pand Submodule is responsible for producinga complete set of cards. Each citation is ex-panded by as many index terms as are con-tained within the entry.
GRACE catalog card records are for-matted on photographic paper for each cardimage, in 6-point upper- and lowercase font.The paper is then developed, inspected, andmicrofilmed. After developing the microfilm,the citations are enlarged to 8-point size andreproduced on the Xerox Copyflo machine.The cards are then cut to size, ho:3-punched,
1/4
COMPUTER-AIDED CATALOG SYSTEM FLOW CHART
NEW
BOOKSDESCRIPTIVE
AND& SUBJECT
JOURNAL CATALOGING
TITLES
\IN CARDS
CARD ..
STOCK
All,..0*-4.-.. ON -4.........
COPYFLOW
CUT &CONTINUOUS REPRODUCE
DRILL HOLES "4 "--
CATALOG
MASTER.WORKSHEET
INPUTPUNCHING
TYPING
HARD
COPYPROOF.
READ
MEDLARS
DICTIONARYTAPE
PAPER
TAPE
MEDLARS
INPUT
MODULE
CUMULATED
CATALOG
TAPE
NEW
TRANSACTIONS
INPUTERRORS
DAILYCATALOGPROCESSING
UPDATED
CUMULATED
CATALOG
TAPE
2011.111.111 .P.M11711
CARD
PRINTTAPE
ICATALOGCARD
SETS
HAND
FILE INNLM CARD
CATALOG
FIGURE 32.
GRACE
CATALOG
CARD IMAGES
(ON GRACE PAPER)
PHOTOGRAPH
WITH
MICROFILM
CAMERA
DAILY INPUTCATALOG CARD
PROCESSING
65
COMPUTEWAIDED CATALOG SYSTEM FLOW CHART
0
BOOK
CATALOG
PROCESSING
BOOK CATALOGPROCESSING
GRACE
(BIWEEKLY,
QUARTERLY,ANNUAL)
PAPER
MASTERS
.1=111110,
PLATEMAKING, (COMMERCIAL
PRINTING, PRINTER)
BINDING
BOOK
CATALOG
MAIL TOSUBSCRIBERS
CORRECTIONS
TO CATALOG
ENTRIES
FILEMAINTENANCE
66
PUNCH &
SIGHT
VERIFY
CORRECTION
TAPES
CUMULATED
CATALOG
TAPE
FIGURE 32a.
FILEMAINTENANCE
PROGRAMS
CORRECTED
CUMULATED
CATALOG
TAPE
4// MEDLARS ZZZ
non :MEDLARS E xxx
la HMO only
20 Cards Only3r0 New Serial Title
4 0 Added Card Set
0 0 Open SerialC 0Closed SerialM EiDMonograph
R °MEDLARS Retypes
IMPRINT(Year Only)
1889
CATALOGER ee2=.0.
2
OAT E26oco
C/
Title Entry
A '. ;:.1.; 3 01 9 '.
CAL Z 695 C991r 1889L
A/
Author and/orTranslated or Eng.Title (MEDLARS &All Oriental) &Lang. Abbr.
"Cutter, CharlesCharles Ansi, 1837-1903.:Rules for a dictionary catalogue.
,,,V/
Vernacular orTransliteratedTitle (for non -English Titles)
.,..,
''MeSH SubjectHeadings
.. . ,,
2d ed. with corrections and additions.
X/ . _
EditionsStatement
.h HoldingsStatement
ImprintWashington,=Govt. Print. Off.,=1889.
Collation 133 p.--- illus. pats. plates
Series Note1J. S.Bureau of Education.Special report on public libraries, pt. 2
Drop Notes
_,
Added Entries(Title and SeriesIncluded)
.
=T
=e Price
x-reference
Dashed on Entries
"Have" Notes/Dispositioncopy 1 6y.
PHS- 3114.1 (REV. 9.651 CATALOG MASTER COMPUTER
FIGURE 33.Sample Specim n Catalog Master (Input Data Sheet).
67
and returned to the Catalog Mainte-nance Section for manual card finishing andfiling in the Library's card catalog.
To produce a biweekly list, the codedequivalents for subject headings are con-verted back to natural language. The cita-tions are expanded by the number of addedentries and subject headings in each record.Only those citations flagged for the desig-nated two-week period appear in any one bi-weekly NLM Current Catalog. The biweeklylist is printed onto photographic film or paperwhich is developed, cut, inspected, and for-warded to a commercial printer, workingunder contract to the Government PrintingOffice. These biweekly issues of NLM Cur-rent Catalog are selective in coverage; thatis, older, non-current materials are not in-cluded.
To produce a quarterly cumulated catalog,the coded equivalents for subject headingsare converted, as for the biweekly list, toEnglish subject headings. Citations are ex-panded for added entries and for authorcross-references, and by subject headings.The quarterly cumulated lists include all ma-terials cataloged by NLM regardless of im-print date, and are printed in the same man-ner as the biweekly issues.
As the catalog cards, the biweekly list, orone of the cumulated catalogs are inspected,errors may be detected. Any correctionsneeded are noted on correction data sheets.Corrections are made on the cards them-selves, as was done in the past, or a newcard set can be generated. In either case, themachine record in the cumulated catalog fileis corrected.
The Library is in the process of convert-ing part of the earlier catalog (pre-1966) tomachine-readable form. Conversion of the1965 catalog is underway, and considerationis being given to converting all materialfrom 1960 (when MeSH was first used forcataloging) to 1964 as well. Availability of
68
a machine-stored catalog will facilitate fu-ture systems for computer-aided acquisitions,searching, and reference activities.
Operating ExperienceThe computer-based cataloging system has
been in operation since January 1966. Oper-ating statistics for the first 24 months areshown in Table 8.
Throughput time for the catalog card isthree days, measured from the time the cat-alog master is forwarded to the Input Sectionto the return of the catalog cards to theCatalog Maintenance Section. Throughputtime for the biweekly is two work weeks,measured from the time when the NLM Cur-rent Catalog is produced by the computer,through the printing and mailing stages bythe Government Printing Office. Throughputtime for the cumulations varies from 17workdays for th.e first quarterly to 43 work-days for the annual.
Table 8Operating Statistics
Item Statistics1966 1967
Number of books cataloged
Number of catalog cardsproduced
12,472 15,744
115,850 156,213
Some operating problems were encoun-tered. Probably the two most significant were(1) input errors occurring more frequentlythan estimated, requiring complete galleyproofreading of each issue; and (2) greatattention to detail required by the catalogerin preparing the input data sheet, which hasincreased cataloging time.
Chapter IX
SUMMARY
This report presents the final chapter onthe current MEDLARS system at the pointin time when the National Library of Medi-cine is on the verge of developing a new,more powerful approach to the problems ofliterature control and information retrievalin the biomedical sciences. This publication,together with the final report of the MED-LARS evaluation project* and the originalMEDLARS story, presents a complete chron-icle of the design and operation of one ofthe world's largest information retrievalsystems.
MEDLARS was designed and operatedaccording to conservative, well-establishedprinciples of documentation and informa-tion science. The most important of thesewere (1) the use of a controlled vocabularyfor indexing and file searching; (2) the useof trained, professional indexers; and (3)serial organization of the search file on mag-netic tape. The Library did not attempt toexperiment with untested theories in design-ing the system, because of the large volumeof literature to be processed and the numberof users to be served. The conservative ap-proach has resulted in four years of contin-ued successful operation. What is more, thereis no evidence of the system breaking downas the volume of literature increases, assome in this field predicted.
Many changes were made to MEDLARSduring its four years of operation. The sys-tem was not static, and the major weak-nesses that were discovered required system
*F. W. Lancaster, Evaluation of the MEDLARSDemand Search Service, National Library of Medi-cine, Bethesda, Maryland, 1968.
modifications. Some of the more significantchanges included:
(1) Change of the basic information storefrom a two-file (one for publication,the other for demand search) to asingle-file approach.
(2) Incorporation of topical subheadingswhich originally were not thought tobe useful for this computer-based sys-tem.
(3) Change of the demand search inputfrom punched paper tape to pre-punched cards.
(4) Major revisions of the Demand Searchand File Maintenance computer mod-ules to improve operating efficiency.
MEDLARS is the only large-scale opera-ting information retrieval system that hassubjected itself to a rigorous, formal evalua-tion. This evaluation showed that the systemwas generally working well at a middle-of-the-road performance level in terms of recalland precision. Many important recommenda-tions for improvement resulted from theevaluation. These included:
(1) User/system interaction needs to beimproved; particularly important is anew and better Search Request Form.
(2) More use should be made of indexersand searchers in updating of MeSH,and less emphasis should be placed onoutside experts.
(3) A better entry vocabulary is neededto allow effective selection of theproper MeSH terms in indexing andsearching.
(4) Expanded use should be made of sub-headings.
(5) Wherever applicable, use pre-estab-lished search strategies for conceptsthat re-occur in searching.
69
(6) The distinction between "depth" and"non-depth" journals for indexingshould be eliminated.
Some of the Library's original objectiveswere met completely; others were not. Ingeneral, however, NLM was able to designand implement a large-volume productionsystem capable of meeting the multiple re-quirements of published indexes and indi-
vidual literature searches. Some of theoriginal plans proved to be unrealisticforexample, the desired ability to provide two-day turnaround for demand searches.
The benefits of operating a system such as
70
MEDLARS are many. Of particular impor-tance is the experience gained from actuallyfacing day-to-day operating :problems. Thepersonnel involved learn more each day andare much. better qualified to move ahead inthe design of more powerful new systemsemploying advanced techniques. Such is thesituation in the Library today. The valuablelessons learned in the operation of MED-LARS are being applied to the design of anew system whose capabilities should be
even more dramatic than those presentedabove. This report, therefore, serves both asa finale for the present MEDLARS and anoverture for the exciting days ahead.
SELECTED REFERENCES
ADAMS, S. Bibliographic organiza-tion in the bioMedical sciences.Wilson Libr Bull 40:714-8, April
1966.. Medical library resources
and their development. J MedEduc 38:20-7, January 1963. Al-
so in: Libr J 88: 2615-22, July1963, as: Medical libraries arein trouble.
. MEDLARS: performance,problems, possibilities. Bull Med
Libr Ass 53:139-51, April 1965.
. MEDLARS: progress and
prospects. In: Rubinoff, M., ed.Toward a National InformationSystem. Washington, Spartan,1965. p. 77-85.
. MEDLARS and the li-
brary community. Bull Med LibrAss 52:171-7, January 1964.
. The MEDLARS system.Fed Proc 22:1018-21, July-Aug-ust 1963.
. More on MEDLARS. Re-
ply to R. R. Shaw's social dis-order. Libr J 90:2450 passim,June 1, 1965.
. The National Library ofMedicine's role in biomedicalcommunications. Bio Sci 15:585-
8, September 1965.. The scientific revolution
and the research library. LibrResour Tech Sery 9:133 -42,
Spring 1965. Also in: Biblos 16:1-24, April 1965.
SEWELL, W. Indexing the
dental literature; MEDLARS.Proceedings, International Con-
ference on Journalism, San
Francisco, Calif., November 6,1964. p. 50-6.
, TAINE, S. I. The NationalLibrary of Medicine and MED-LARS. Rev lnt Doc 31:107-10,
August 1964.
--. Searching the med-ical literature. Information re-trieval (MEDLARS) at the Na-tional Library of Medicine.JAMA 188:251-4, April 20, 1964.
ATWOOD, R. A grass-roots look atMEDLARS. Bull Med Libr Ass52:645-51, October 1964.
AUSTIN, C. J. The computer andmedical bibliography. Milit Med129:737-41, August 1964.
. Data processing aspectsof MEDLARS. Bull Med LibrAss 52:159-63, January 1964.
: Dissemination of informa-tion and problems of graphicpresentation. International Fed-eration for Documentation. Pro-ceedings of the 31st Meeting andCongress, Washington, D.C., Oc-
tober 1965. Washington, Spar-tan, 1966. p. 1241-5.
. Experience with theGRACE system. In: Hattery, L.H. and Bush, G..P., 'eds.. Automa-
tion and Electronics in Publish-ing. Washington, Spartan, 1965.
p. 61-70.The MEDLARS project
at the National Library of Medi-cine. Libr Resour Tech Sery 9:94-9, Winter 1965.
. The MEDLARS system:an application report. Datama-tion 10:28-31, December 1964.
BARNETT, M. P. Computer Type-setting; Experiments and Pros-pects. Cambridge, M.I.T. Press,1965. p. 166, 180-5, 201, 227.
BECKER, J. The MEDLARS proj-ect. ALA Bull 58:277-30, March
1964.BUCK, C. A. Adverse reactions to
drugs and drug recall systems
from the IPAD viewpoint. Milit
Med 130:49'7-501, May 1965.
BLOOMQUIST, H. The status sndneeds of medical school librariesin the United States. J MedEduc 38:145-63, March 1968.
BOGART, R. E. Use of GRACE atNLIVI. In: Lee, R. W. and War-ral, R. W., eds. Proceedings of aSymposium, National Bureau ofStandards, June 15-18, 1807.,
(NBS Special Publication 295,February 1968). p. 94-7.
BRODMAN, E. Machines in medi-cal libraries. Med Bid Illus 16:238-42, October 1966.
CHAREN, T. MEDLARS pro-gram of the National Libraryof Medicine; indexing operation.Presented at the 3d Annual In-stitute on Technical Literature.Abstracting and Indexing. TheAmerican University --Centerfor Technology and Adminis-tration, Washington, D.C., May15-17, 1967. To be published.
CLAPP, V. W. Mechanization indautomation in American librar-ies. Libri 14:369-75, 1964.
COBLANS, H. Use of MechanizedMethods in Documentation Work.London, Aslib, 1966. 89p.
COOLEY, S. G. Medical bibliogra-phy and MEDLARS. Mast Re-constr Sur, 40:113 -6, August1967.
COMPUTER USAGE CO., INC. Con-version of NIH Data. Baltimore,Md., 1965. 2:1p. (AD-621066)
CUADRA, C. A. Annual Review ofInformation Science and Teckwi-ogy, v. 1. New York, Intereci-ence, 1966. p. 119, 179, 295-7,316.
CUMMINGS, M. M. The biomedi-cal communications problem.
Ciba Foundation Symposium onCommunication in Science: Doc-umentation and Automation,London, Churchill, 1966. p. 110-
22.
71
-, Books and computers(from Billings to MEDLARS).Acad Med N J Bull 11:51-6,March 1965.
Health sciences (MED-LARS). In: Kent, A. and Taul-bee, 0., ed. Electronic Informa-tion Handling. Washington, Spar-tan, 1965. p. 111-9.
. A library-based computersystem for indexing, storing andretrieving literature in the car-diovascular field. Ann N Y AcadSci 126:919-25, August 6, 1965.
. MEDLARS in operation.Comment on Shaw, R. R.: Formand the substance. Libr J 90:1218, March 15, 1965.
. The National Library ofMedicine: new program in sup-port of medical education, re-search, and practice. Clin Anesth1:115-23, 1966.
. The National Library ofMedicine and the practitioner.Postgrad Med 39:327-8, March1966.
. Needs in the health sci-ences. Proc. of Electronic Han-dling of Info Testing & Evalua-tion. Washington, Thompson,1967.
. Plans for the develop-ment of a medical library net-work. In: Simonton, W. and Ma-son, C., eds. Information Retriev-al with Special Reference to theBiomedical Sciences; papers pre-sented at the 2d Institute onInformation Retrieval conductedby the Library School and theNolte Center for Continuing Ed-ucation, University of Minneso-ta, November 10-13, 1965. Min-neapolis, Minn., Nolte Centerfor Continuing Education, Uni-versity of Minnesota, 1966. p.17542.
. Scope and services of theU.S. National Library of Medi-cine. Israel j Med Sci 1:180-4,March 1965.
DARLING, L. Information retrievalprojects in the Biomedical Li-brary, University of California,Los Angeles. Presented at theClinic on Library Applications ofData Processing, University ofIllinois, Urbana, Ill., April 24-27,
72
1966. To be published by U. Ill.Press.
. MEDLARS: a regionalsearch center. In: Simonton, W.and Mason, C., eds. Op. cit. p.
49-62.. The MEDLARS regional
center at UCLA: a progress re-port. Bull Med Libr Ass 64:311-5, October 1966.
. Progress report on theMEDLARS project: opening re-marks. Bull Med Libr Ass 52:148-9, January 1964.
DAVIS, A. C., BOOTH, F. E. Pro-gramming and coding techniquesused in MEDLARS. In: Luhn,H. P. and Janaske, P. C., ed. Au-tomation and Scientific Commu-nication; Short Papers . . . 26thAnnual Meeting of the Ameri-can Documentation Institute,Chicago, Ill., October 6-11, 1963.Washington, 1963. p. 223-4.
DETERLING, R. A., JR. The role ofplastics and engineering in vas-cular surgery. Arch Surg (Chi-cago) 93:697 -703, November1966..
Diverr, R. T. Annual administra-tive reviews: library service.Hospitals 39:107-10, April 1,
1965.ELWIN, C.-E. The use of MED-
LARS by the pharmacologist:some aspects of terminologicproblems. Pharmacologist 8:End. 18, Summer 1966.
ESTERQUEST, R. T. Improving thequality of the medical school li-
brary. J Med Educ 39:451-8,May 1964.
. The program of the Fran-cis A. Countway Library of Medi-cine. Bull Med Libr Ass 55:38-41, January 1967.
FERNSTROM, R. (Pharmaceutic Aand 0 by OA and electronic dataprocessing) Svensk Farm T 70:
267-72, April 10, 1966. (Sw)
FLOOD, B. Analysis, of questionsaddressed to a medical referenceretrieval system: Comparison ofquestion and system terminolo-gies. Amer Doc 18:216-27, Octo-ber 1967.
FRANCRE, G., SEWELL, W. TheDrug Literature Program of theNational Library of Medicineand its relationship to hospital-
based drug information needs.Presented at the 2d AnnualMeeting of the Drug Informa-tion Association, Chicago, Ill.,June 25-26, 1966. Unpublished;obtainable from DLP, NLM.
FRHS, H. (Medical informationretrieval) Ugeskr Laeg 127:1446-7, November 11, 1965.(Dan)
GARRARD, R. F. MEDLARS: Sys-tems engineering applied to li-braries. In: Simonton, W. C., ed.Information Retrieval Today;papers presented at the Instituteconducted by the Library Schooland the Center for ContinuationStudy, University of Minnesota,1963. p. 119-39.
. Misleading and unfound-ed. Comment on Shaw, R. R.:Form and the substance. Libr I90:15.80, April 1, 1965.
GARviS, F. J. The UCLA MED-LARS computer system. BullMed Libr Ass 54:14-5, January1966.
GERBER, L. S. An automated sys-tem for handling biomedical lit-erature. J Amer Vet Med Ass147:1530-5, December- 15, 1965.
GIBBON, J. H., JR. A note on thePhiladelphia Regional MedicalLibrary Committee. Bull MedLibr Ass 53:559-62, October1965.
GOLDWYN, A. J. The doctor andthe document: a new tool forbiomedical research. Wilson LibrBull 36:829-32 passim, June1962.
HARLEY, A. J., BARRACLOUGH, E. D.MEDLARS information retriev-al in Britain. Postgrad Med J42:69-73, February 1966.
HENDERSON, M. M., MOATS, J. S.,STEVENS, M. E., NEWMAN, S.Cooperation, Convertibility, andCompatibility Among Informa-tion Systems: A Literature Re-view. Washington, National Bu-reau of Standards, 1966. 140 p.U.S. National Bureau of Stand-ards. Miscellaneous Publication,276)(BILLINGS, J. S.), HILL, L. The
medical library crisis: Billingsto MEDLARS. J Med Educ 41:80-4, January 1966.
HOGAN, R. An Evaluation ofMEDLARS output: demand and
recurring bibliographies. BullMed Libr Ass 54:321-4, October1966.
HOLZBATJER, H. Bibliographic or-ganization in the Federal Gov-ernment. Wilson Libr Bull 40:719-20 passim, April 1966.
HOWARD, E. Mechanization inmedical libraries. South DakotaJ Med Pharm 16:31-2, June1963.
HUTH, E. J. Out of computer byjoint: Index of Rheumatology.Ann Intern Med 62:626, March
1965.INSTITUTE FOR ADVANCEMENT OF
MEDICAL COMMUNICATION. DrugInformation for the BiomedicalCommunity; a report of prelim-inary study of the needs for anational drug information clear-inghouse, by William C. Springand F. Honicker. Bethesda, Md.,1963. p. 66-8.
Izzo, J L. Identification of theliterature of a specialized sub-
ject. In: Simonton,. W. and Ma-son, C., eds. Op. cit. p. 63-68.
KAREL, L. The physician and theNational Library of Medicine.Penna Med 1967, 70, 1964.
, AUSTIN, C. J., CUMMINGS,
M. M. Computerized biblio-graphic services for biomedicine.Science 148:766-72, May 7, 1965.
KENT, A. Computers and biomedi-
cal information storage and re-trieval. JAM:' 196:927-32, June
1966.
KENTON, C. MEDLARSprogramof the National Library of
Medicine; searching operation.Presented at the 3d Annual In-stitute on Technical LiteratureAbstracting and Indexing. The
American UniversityCenter forTechnology and Administration,Washington, D.C., May 15-17,
1967. To be published.. The recurring bibliogra-
phies program of MEDLARS.Bull Med Libr Ass 54:135-7,April 1966.
LANCASTER, F. W. Evaluating theperformance of a large operatingretrieval system. Proc of Elec-tronic Handling of Info Test-ing & Evaluation. Washington,Thompson, 1967.
. Evaluation of the MED-
LARS Demand Search Service.National Library of Medicine,Bethesda, Md., 1968.
LANGNER, M. C. The National Li-brary of Medicine's services: ini-tial observations on indexing, in-terlibrary loan, and reference.Bull Med Libr Ass 55:31-7, Jan-uary 1967.
LEMAN, E. National Library ofMedicine: a library's library.,Hosp Top 44:59-62 passim, Aug-ust 1966.
LEWIS, R. F. The MEDLARSproject at the UCLA BiomedicalLibrary. Bull Med Libr Ass 54:11-3, January 1966.
LUIGART, F. W. Push-button brainsproduce faster facts for doc-
tors; the National Library ofMedicine collects and dispensesknowledge electronically. (Wash-
ington, Courier-Journal, 1964)
One section of the Courier-Jour-nal, Louisville, Ky., November22, 1964. p. 25-32.
LUSTED, L. B. Computers andmedical data processing. Port-land Clin Bull 15:31-8, March
1964.LYNN, K. C. The computerized
analysis, storage, and retrievalof dental literature. Presentedat the Combined Public HealthService Commissioned Officers
Association-Clinical SocietyMeeting, Baltimore, Md., May
12, 1966. Unpublished; requestfrom author at NLM.
MCCANN, A. Advantages of a uni-versal coding and classificationsystem for drugs: implications ofclassification for medical subjectheadings. Am J Hosp Pharm 23:
87-8, February 1966., CLEVELAND, C., ELWIN, C.-
E. American Society of Hospi-tal PharmacistsNational Li-
brary of Medicine collaborationin standardizing class terms fordrugs for mechanized searching.Proc Ann Meeting Drug InformAss 2:35-48, 1967.
MANN, G. (On the situation of
medical librarianship in the Fed-eral Republic of Germany) MedWelt 14:796-802, April 4, 1964.(Ger)
MARTIN, E. W. (The medical in-formation system in the USA)
Svensk Farm T 70:351-8, May
20, 1966; 37.5-86, May 30, 1966.
(Sw)MOLL, W. The medical library
and the scientific revolution. Vir-ginia Med Monthly 91:27-30,January 1964.
MORTON, L. T. Development of aregional MEDLARS station inBritain. Bull Med Libr Ass 53:448, July 1965.
OHTA, M. A comparison of somedemand subject searches: ma-chine vs. human. Bull Med LibrAss 55:408-15, October 1967.
OLcH, P. Medical subject head-ings (MeSH). Proceedings of theAMA Workshop on BiomedicalVocabulary Control, October 30-31, 1964. Chicago, 1965.
. New approaches to utili-zation of library resources incontinuing education at the Na-tional Library of Medicine. In:Russell, W. 0. and Kolvoord, R.
A., eds. A Symposium: Implica-tions of Developments andTrends in Educational Technol-
ogy Related to the ContinuingEducation of Physicians. Chica-
go, American Society of ClinicalPathologists, 1966. p. 96-9.
ORR, R. H., PINGS, V. M. Docu-ment retrieval: the national bio-medical library system and in-terlibrary loans. Fed Proc 23:
1155-63, September-October 1964.
, Limps, A. A. Ref-erence retrieval tools: biomedi-cal abstracting and indexingservices. Fed Proc 23:1164-76,September-October 1964.
PINGS, V. M. A Plan for Indexingthe Periodical Literature ofNursing; report of a study of theneed for bibliographic control ofnursing. New York, AmericanNurses' Foundation, 1966. 202p.
RANNIE, I. Medical literature re-trieval. Brit Med J 5499:1351-2,May 28, 1966.
RAVDIN, I. S. The importance ofthe library in medical educa-tion. Bull Med Libr Ass 53:505-9, October 1965.
REDMER, P. Publication by GRACE.Meth Inform Med 4:145-50, Sep-tember 1965.
ROBERTS, A. E. A librarian's visit
r70 04
to America. Bristol Medicochir J80:19-22, April 1965.
ROGERS, F. B. Changes in Medi-cal Subject Headings (Commu-nication to the Editor) BullMed Libr Ass 51:114-6, January1963.
--,---. The development of MED-LARS. Bull Med Libr Ass 52:1.;11-1, January 1964.
MEDLARS operating ex-1.erience at the University ofColorado. Bull Med Libr Ass 54:140, January 1966.
--,. MEDLARS operating ex-Nrier.;e: addendum. Bull Med
ihr Ass 54:316-20, October1966.
---. The National Library ofMedicine's role in improvingmedical communications. Amer Jilled Electronics 1:230-41, July->3.-43 tember 1962.
Stresses in current medi-cal bibliography. New Eng JMed 267:704-8, October 4, 1962.
RUHL, M. J. Development of amedical specialty recurring bibli-ographyIndex of Rheumatolo-gy. Bull Med Libr Ass 55:70-4,January 1967.
, SOKOLOFF, L. The litera-ture analysis project of theAmerican Rheumatism Associa-tion. Arthritis Rheum 7:615-22,December 1964.
SADUSK, J. F., JR. Reflectionsupon biomedical communications.Drug Inf Bull 1:4-10, October-December 1966.
SCHILLER, H. What is MED-LARS? Libr J 88:949-53, March1, 1963..
SCHMIDT, A. (MEDLARS) NordMed 72:1268-9, October 22,1964. (Dan)
SEWELL, W. Automation in index-ingMEDLARS in medicine.J Amer Pharm Ass NS 4:292-4,June 1964.
. The impact of the Na-tional Library of Medicine onthe small information group.J Chem Doc 6:168-70, August1966.
. Medical subject headingsin MEDLARS. Bull Med LibrAss 52:164-70, January 1964.
. The National Library ofMedicine's drug literature pro-
74
gram. Proc Ann Meeting DrugInform Ass 2:1-8, 1967.
SHAW, It. R. The form and thesubstance. Libr J 90:567-71,February 1, 1965.
. Social disorder (Rejoin-der to Cummings, M. M.: MED-LARS in operation). Libr J 90:1580 passim, April 1, 1965.
SHERA, J. H., BRODMAN, E.Guidelines for medical school li-braries: machine retrieval sys-tems and automated proceduresin libraries. J Med Educ 40:46-51, January 1965.
SHUMWAY, N. MEDLARS: vo-cabulary construction and medi-cal subject headings. In: Simon-ton, W. and Mason, C., eds.Op. cit. p. 35-47.
SJOBERG, B. (Drug documentationon international, industrial andhospital level). Farm Revy 65:614-21, October 5, 1966. (Sw)
SMITH, J. P. Aids to retrieval ofinformation. Milit Med 129:774-6, August 1964.
SNYDER, J. D. "Drug LiteratureProgram" offers hospitals world-wide data. Hosp Form Manage2:46-8, March 1967.
STEAD, E. A. Your National Li-brary of Medicine. Med Times94:507-8, April 1966.
STEUDEL, J. (Medical school li-braries and medical bibliographyin the USA) Deutsch Med TV schr88:1957-60, October 4, 1963.(Ger)
TAINE, S. I. Bibliographic aspectsof MEDLARS. Bull Med LibrAss 52:152-8, January 1964.
. Bibliographic data proc-essing at the National Libraryof Medicine. In: Goldhor, H., ed.Proceedings of the 1963 Clinicon Library Applications of DataProcessing, University of Illi-nois, Urbana, Ill., April 28-May1, 1963. Ann Arbor, Mich., Ed-wards, 1964. p. 109-32.
. The future of the pub-lished index. In: Machine Index-ing; Progress and Problems. Pa-pers presented at the 3d Insti-tute on Information Storage andRetrieval, American University-Center for Technology and Ad-ministration, Washington, D.C.,February 13-17, 1961. (Wash-ington, 1962) p. 144-69.
. The medical literatureanalysis and retrieval system.Bull Med Libr Ass 51:157-67,April 1963.
. The medical literatureanalysis and retrieval system(MEDLARS) of the U.S. Na-tional Library of Medicine. MethInform Med 2:65-9, April 1963.
. The National Library ofMedicine medical literature an-alysis and retrieval system(MEDLARS). In: Luhn, H. P.and Janaske, P. C., eds. Op. cit.p. 105-6.
TANIGUCHI, M. (Demand searchservice for MEDLARS) Igakutoshokan 11:201-17, 1964. (Jap)
TAUBE, M. The coming of age ofinformation technology. Bull MedLib?. Ass 52:120-7, January1964.
TERRY, L. L. Biomedical commu-nications in the public healthservice. Amer J Hosp Pharm23:304-6, June 1966.
. The crisis in health com-munications. Hospitals 38:49-51,June 16, 1964.
TRUELSON, S. D., JR. What theIndex Medicus indexes, and why.Bull Med Libr Ass 54:329-36,October 1966.
U.S. NATIONAL LIBRARY OF MEDI-CINE. Drug literature; a fac-tual survey on "The nature andmagnitude of drug literature."Report prepared for the studyof "Interagency Coordination inDrug Research and Regulation"by the Subcommittee on Reor-ganization and International Or-ganizations of Senate Commit-tee on Government Operations.(Washington) 1963. p. IX, 20, 51,160-3.
. The MEDLARS Stcry atthe National Library of Medi-cine. (Washington, 1963) 74p.
. National Library of Medi-cine. (Bethesda, Md., 1966) p.4-5. (Public Health Service Pub-lication No. 1533)
. Guide to MEDLARS Serv-ices. Bethesda, Md., 1967. 22p.(Public Health Service Publica-tion No. 1694)
U.S. PRESIDENT'S COMMISSION ONHEART DISEASE, CANCER ANDSTROKE. Report to the Presi-dent; a national program to con-
:1
quer heart disease, cancer, andstroke. (Washington) 1964-1965;v. 1:24-5, 64-6; v. 2:396-9.
U.S. PUBLIC HEALTH SERVICE. OF-FICE OF THE SURGEON GENERAL.Organization study: National Li-brary of Medicine. (Washing-ton) 1965. 48p.
WAGNER, G. (On the informationproblem in the area of medicine)Heidelberger Jahrbucher 9:39-53, 1965. ;Ger)
WASHINGTON, 1.1, The auxiliarychemical module MEDLARS'"switching station" for drug lit-erature. Proc Ann Meeting DrugInform Ass 2:49-58, 1967.
WEIss, I. J. Electronic informa-tion processing in a modern li-brary. Presented at the WaterResources Engineering Confer-ence, American Society of CivilEngineers, Mobile, Ala., March8-12, 1965.
, WIGGINS, E. Computer-aided centralized cataloging atthe National Library of Medi-cine. Libr Resour Tech Sem11:83-96, Winter 1967.
WIEDLING, S. (New documenta-tion system) Farm Revy 65:593-604, September 25, 1966. (Sw)
WILSON, M. P., CUMMINGS, M. M.The National Library of Medi-cine: relationships to medicaleducation and research. J MedEduc 40:225-32, March 1965.
, DOUGLASS, C. D., KEFAU-VER, D. F. Extramural pro-grams of the Natonal Library ofMedicine: program objectivesand present status. Bull MedLibr Ass 54:293-310, October1966.
YAST, H. T. MEDLARS: its im-plications for hospital libraries.Cath, Libr World 34:488 passim,May 1963.
ANONYMOUSASLIB PROC. Experiment in medi-
cal information. As lib Proc 17:180, June 1965.
. U.K. MEDLARS in oper-ation. As lib Proc 18:149, June1966.
BIOMED INSTRUM. Computer ap-plications: MEDLARS. BiomedInstrum 1:22-4, October 1964.
BRIT MED J. Using medical li-
braries. Brit Med J 1:1483, June11, 1966.
. Improving medical com-munication. Brit Med J 1:517-8,March 4, 1967.
BULL MED LIBR Ass. MEDLARSproject begins. Bull Med LibrAss 50:120, January 1962.
. NLM MEDLARS sympo-sium. Bull Med Libr Ass 51:438-9, July 1963.
. Nation's speediest photo-typesetter installed at NLM.Bull Med Libr Ass 52:801-2,October 1964.
. Pilot study for MED-LARS decentralization. Bull MedLibr Ass 53:126-7, January1965.
Time capsule to featureNLM and MEDLARS. Bull MedLibr Ass 53:275-6, April 1965.
. British library official be-gins MEDLARS training. BullMed Libr Ass 53:277, April 1965.
MEDLARS to generateIndex to Dental Literature. BullMed Libr Ass 53:281, April 1965.
. New MEDLARS centerat Colorado. Bull Med Libr Ass53:473-4, July 1965.
International Nursing In-dex to be produced by NLM andAJN (American Journal ofNursing) Company. Bull MedLibr Ass 53:474, July 1965.
. New MEDLARS centersat Harvard and Michigan. BullMed Libr Ass 54:272, July 1966.
. Overseas MEDLARS cen-ter. Bull Med Libr Ass 54:442-3, October 1966.
. MEDLARS to expand.Bull Med Libr Ass 55:105, Jan-uary 1967.
NLM committee on se-lection of literature for MED-LARS. Bull Med Libr Ass 55:237, April 1967.
DATA PROCESSING. Medical litera-ture analysis and retrieval sys-tem. Data Processing 7:58-60,January-February 1965.
FED PROC. Communication prob-lems in biochemical research: re-port of a study. Fed Proc 23:1117-32, September-October 1964.
HOSPITALS. Decentralizing MED-LARS is aim of UCLA study.Hospitals 39:129 passim, Janu-ary 16, 1965.
JAMA. Cumulated Index Medicus.JAMA 193:130-1, July 19, 1965.
J AMER VET MED Ass. The Na-tional Library of Medicine. JAmer Vet Med Ass 147:473-5,September 1, 1965.
J MED Ass ALABAMA. Medicalcenter named MEDLARS site.J Med Ass Alabama 35:838, Ap-ril 1966.
LAKARTIDNINGEN. (MEDLARSgives quick information) Lakar-tidningen 62:815, March 17,
1965. (Sw)LIBR CONG INF BULL. MEDLARS
(Medical Literature Analysisand Retrieval System). LibrCong Inf Bull 20:612-3, October16, 1961.
. Nation's fastest compu-ter-driven phototypesetter in-stalled as part of MEDLARS.Libr Cong Inf Bull 23:413, Aug-ust 3, 1964.
National Library of Med-icine has awarded contract toUCLA for pilot study in con-nection with proposed decentrali-zation of MEDLARS. Libr CongInf Bull 23:704, December 28,1964.
LIBR J. Electronic retrieval sys-tem for National Library ofMedicine. Libr J 86:3928, No-vember 15, 1961.
. Medical library acquiresGRACE: nation's fastest photo-typesetter. Libr J 89:3122, Sep-tember 1, 1964.
. UCLA gets contract forpilot study on decentralization ofMEDLARS. Libr J 90:603-4,February 1, 1965.
. NLM publications to beburied in World's Fair time cap-sule. Libr J 90:1276, March 15,1965.
. Second regional MED-LARS center to be University ofColorado. Libr J 90:2786 passim,June 15, 1965.
. British experiment withMEDLARS; NLM to providetapes and training. Libr J 90:2990-1, July 1965.
. Route 66A selective re-view and commentary on the(mostly American) library merleand its events of the past year.Libr J 92: 51-63, January 1, 1967.
75 1)7,
NATURE. More from MEDLARS.Nature 213:223, January 21,
1967.NEWSWEEK. The new librarian.
Newsweek (4) 85-6, July 25,1966.
NURS FORUM. A progress note ona nursing index. Nurs Forum3:7, April 1965.
PUB WKLY.. MEDLARS data sys-tem is subject of briefing. PubWkly 187:105.6, January 4,
1965.REV Asoc ODONT ARGENT. (A new
76
stage of the "Index to DentalLiterature") Rev Asoc Odont Ar-gent 54:227-8, June 1966. (Sp)
Sci INF NoTEs. MEDLARS andGRACE team-up to publish Aug-ust issue of NLM's Index Medi-cus. Sci Inf Notes 6:9-10, Aug-ust-September 1964.
. MEDLARS decentraliza-tion subject of UCLA study. SciInf Notes 7:14, February-March1965.
SCI NEWS LETTER. U.S. service----cuts British research time; corn-
puter-operated information re-trieval service of medical litera-ture. Sci News Letter 90:339,October 22, 1966.
SPEC LIBR. Alabama MEDLARSsearch center. Spec Libr 57:197,March 1966.
. Further MEDLARSstudy. Spec Libr 57:721, Decem-ber 1966.
WILSON LIBR Bum,. MEDLARSinstalls GRACE. Wilson. LibrBull 39:24, September 1964.
* U.S. Government Printing Office: 1968 0 308-816
Public Health Service Publication No. 1823
U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE
Public Health Service
National Institutes of Health
National Library of Medicine
