ARCHIVAL INFORMATICSNEWSLETTER

Part.l of Archival Infonnatics Newsletter & Technical Reports

ISSN 0892·2179

FROM THE EDITOR:The fIrst issue of any new publication calls

for explanation. The simplest one is almostcertainly the best - I felt a real need amongarchiVists to discuss automation and its impacton their profession and I recognized a lack ofvehicles for this dialog. After leaving theSmithsonian last fall. I found myself building adatabase of archival automation news which Iknew had not been available to me when I wasDeputy Director for Infonnation ResourceManagement at a very large, publicorganization. It was obvious to me that it wasnot available to others either. Then I fmUldhow many terrific applications have neverbeen reported, how many studies, such as thereport on archiving of Database ManagementSystems by Tom Brown and Bill Reader in thisissue, cannot find their way to print, and howmany thoughtful discussions, like BarbaraCain's treatment of indexing issues in the NorthCarolina State Archives system manual areburied beyond the reach of colleagial review. Ihope the Archival InfQnnalics Newsletterattracts your good ideas and stimulates avigorous dialog. As always, I will provide lots ofmy own ideas to shoot at! Please send insnippets Qr diatribes, reviews or news of yourown.

But anewsletter cannot support some of theneeds of the community; some extremelyimportant technologies need to be examined ingreater deatail, and reported from anarchivist's perspective for the use of archivesadministrators. This issue of the ArchivalInfQrmatics Newsletter is part 1 of ArchivalInformatics Newsletter & Technical Re.port:s.This quarter, part 2 is an extensive technicalreport on optical technologies and theirsignificance for archives today. It is intendedto be extremely practical and directly usable - Ileave to your judgments whether it works, but

Volume 1, #1

fm sure such technology forecasting andpragmatic advice are needed. Future topics forthe Technical Re.ports include archiving ofsoftware, artificial intelligence, requirementsfor automated archival control systems, and theuse ofinfonnation technologies in collectionsmanagement and tracking. There are manyother topics I would like to research myself, butI would also like to invite others with specialareas of technical knowledge to propose futurestudies and author them as part of theTechnical Report series;

David Bearman, Editor

TABLE OF CONTENTS:

ColumnsAutomated System: Views 2-4

by David BearmanMachine Readable: Views 5-7

by Thomas E. Brown

What are/is Infonnatics? 8

Studies &·Surveys:ART Task Force Report on ArchivingData Base Management Systems 9..12

DeUartments:Conferences 13In-Box 14-15Follow-up on the News 16Plans & Proposals : ; 17

Standards 18Technical Report: Optical Technologies 19

AUTOMATED SYSTEMS:by David Bearman

INHERITANCE OF INDEXING TERMSINFAIDS

The North Carolina State Archives hasrecently implemented a system for holdingslocation and retrieval, information retrievalfrom fmding aids and limited administrativeprocessing control functions on the Prime 9655mini-computer it shares with the rest of theDepartment of Cultural Resources. The systemrons on PRIMOS, using Prime Information andBlacksmith Intercept software and an in-housedata entry program. The system uses 34essentially standard data elements and plans toproduce MARC formatted output Currently it isimplemented for staff use as the database(18,158 records with an average length of 205characters in early March) is being builtEventually it is intended to serve as an on-linepublic access catalog and will implementauthority control over subjects consistent withLC headings as far as is possible.

At some time in the future, I hope to reviewthe software, but here I will discuss a numberof issues relating to indexing of archivalmaterial in hierarchically structured fileswhich North Carolina has attempted to resolve.Like many other archival systems, the databasearchitecture implemented by Dr. Arion Kemple,Computer Specialist for the Department, inresponse to requirements defmed by theArchives Committee on Automated Finding Aids," reflects the organization of records bymaintaining an explicit hierarchicalrelationship between record group, series, box,folder and item, linking information enteredfor the higher level with that entered Jor thelower levels".

This method of hierarchical linkage, inwhich all records have the same structureregardless of level but each is assigne,d anidentification number which indicates itsstructural location within the overall holdingsof the institution, is now widely adopted. Thestructural notation is an automated recordrelationship manipulation device only, andbears no relationship to 'call' numbers or stacklocations.

The North Carolina staff identified a varietyof advantages of using such a system in its staffmanual on the system:

" information displayed about a document or

map or photograph at the lowest level willalways include the provenance of thatdocument...

... it also has great flexibility, allowingdescription to whatever level is best for thespecific material -- item level for photographsand maps, box or folder level for most stateagency and county records...

... one can browse the structural arrangementof the Archives...request names of recordgroups or collections that have been entered...alisting of the series in a particular recordgroup, or the boxes in a particular series, orthe folders in a box, or the items in a folder..."

In a section of its documentation entitled'Indexing Rules', the Archives discusses itsapproach to exploiting hierarchicalinheritance:

"Because the archival levels are linked inFAIDS, index entries for material described atone level are always applied also to the lowerlevels contained within that unit. If a personalname, corporate name, geographic name; orsubject is basically involved in all the recordsor papers in a record group or collection; or ina series or a box, even if it is not actuallynamed in every folder, the index entry shouldbe made at the appropriate higher level.Orange County would be entered at the recordgroup level for Orange County records, andAshe, Samuel A'Court would be entered at therecord group level for the Samuel A'Court Ashepapers. If the name or subject is searched foralone, the reference given would be the entirerecord group or collection. If it is entered aspart of a Boolean combination search, it willmatch with index entries made for everycomputer record at every level below it.Orange County + Wills would match at the serieslevel. Orange County -t Wills + John Doe wouldmatch at the folder or item level.

a. In such a system, the computer is bound tocome up with unwanted references from timeto time. If Education and Schools are entered assubjects for the State Board ofEducation Recordgroup, a search for School + Pender County isgoing to produce references to swamp land inHolly Shelter Swamp, since Swamp LandRecords is a series within the Board ofEducation record group. The researcher willusually know from the title of the referencethat the record is not of interest to him, but ifin doubt, he can call up the full description inFAlDS.

b. To reduce the number of inappropriatereferences produced, care should be taken toindex each entry at the best level. For example,

2 Archival Infonnatics Newsletter voU, #1

it may be better to enter 'Schools' at the serieslevel instead of the record group level."

Because the experience of North Carolina is ofgeneral interest and the instructions theyprovide for their indexers are so clear, it isuseful to use their discussion as a point ofdeparture for a critique of index tenninheritance issues in archives.

Although other software environments,especially those used in artificial intelligencesystems employing semantic netwoIks withframe representation (which inherit byexception), can pennit this issue to be finessed,most archives would consider themselvesfortunate to have a system which inheritedthrough the relatively mechanical hierarchydescribed by North Carolina and whichsupported a query which used both assignedand inherited terms. What issues does thisraise?

First, context is always relative. In NorthCarolina. the tenns North Carolina is notentered. but Tennessee is when the recordshave that subject.IfNorth Carolina wasentered. the United States would not be.Inheritance pennits the assignment of broaderterms at higher levels and narrower terms atlower levels, but it does not serve as athesaurus. This is a problem, both because ofwhat we tend not to index (N.C., USA) , andbecause of what it requires us to index. A true

. thesaurus would be a valuable aid here.Second, archivists have a bad habit of

deriving guide entries in which some termsfrom different levels of the description arefeatured in an abbreviated description, notbecause they are important to the collection somuch as for their importance to the audience.Thus, we have many guide entries which indexby Presidents of the United States or byimportant events, even though these arerelatively minor features of the collections. Ifthis practice is deemed unifonnly misleading,using only terms from higher in the hierarchywould be an improvement, but how do weidentify a level from which such tenns will bedrawn so as to distinguish between holdings?

Third, the primary advantage of searchingboth assigned tenns and inherited terms is toreverse the effect of indexing by the mostspecific available entry and assigning suchtenns at the lowest level to which the analysisprocedes; this effect is tenn scattering, whichcan only be allowed in authority controlledindexing with tenn hierarchies. At the sametime, we must be careful !.hat by pennittingassignment of broader terms at higher levels,

many general terms do not rapidly becomeuseless for retrieval because of the numbers ofitems with which they will be associated.

At the core of the quandry is an extremelyvaluable indexing precept - that the narrowestpossible tenn which adequately reflects thecontent of an item should be assigned at alltimes. As North Carolina has discovered,inheritance pennits the archivist to follow thisuseful guideline by assigning narrow tenns atlower levels except in the case of non­homogenous groups (swamps in schoolrecords) when departure from the guidelineagain risks overwhelming the system with toomany occurences of certain tenns.

What the North Carolina system introduces byindexing hierarchical.record levels isequivalent to maintaining·a hierarchicallycontrolled vocabulary so that Holly ShelterSwamp will reference from Orange Countywithout an added explicit reference at eachlevel in the record Nevertheless, they havefound a middle ground which seems viable. Iwould like to invite readers with othersolutions to the overloaded indexing problem,and other methods of supporting narrower andbroader searches to write.

WHY ADOPT MARC?For several years now I have met archivists

who ask me why the archival communityshould adopt MARC. I almost invariably givethree answers which have nothing to do withsharing information. I·would like to repeatthem here, if only to stimulate discussion.

First, we should adopt MARC because it is aninsurance policy for automated data. Otherformats in which we might store the data arenon-standard (wimess the ART Task Forcereport on DBMS's in this issue). If we want to becertain that we will be able to move the data welaboriously compile about our holdings fromone automated environment to another, weneed to be able to read it into and out of aformat which allows for "migration" betweenhardware vendors and software systems.

Second, we should adopt MARC because theAMC fonnat embodies standard archivalpractice, and as a very small profession ourbest chance of being able to leverage sufficientinvestments to get a reasonable variety ofsoftware products from which to selectsomething which meets our needs, is to be ableto provide developers of such products with astandard set of data they must address.

Third, we have an opportunity to disseminate

Spring, 1987 Copyright by Archives & Museum Infonnatics 3

infonnation about our holdings through a vastarray of other agencies simply by being ableto output it in MARC fonnat If we haddeveloped a new food product and wanted toreach the marketplace, we would adoptbarcoding on our packages because the largestgrocery chains would not sell our commodityunless it was barcoded. As archivists, if we wantto reach the infonnation dissemination .marketplace, we need to use the facilities of thenational networks and database vendors. Torefuse to place MARC on our packages is.akin tonot recording barcodes. We will find ourselvesstanding on a comer with a small vendingstand, wondering why we haven't reached alarger audience.

What is noteworthy about this list ofarguments is that none of them hinges on anyassumptions about archivists sharing, or notsharing, data with each other. None of themdepends at all on the realization of .nationalarchival infonnation systems. And bythemselves, none of them.sugg~anythingabout the promise of MARC for cooperativedocumentation or cooperative collectionsmanagement. It is not irrelevant that some ofthe data we want in our archives - like statepublications cataloging or newspapercataloging - is already in MARC fonnats whichwe can't integrate unless we can also readMARC into our systems, but the threearguments aduced before reniam the strongestreasons to adopt MARC, and don't alude to thisreality of the information enviI:onment either!

ARCHIVAL PROJECTS USINGOPTICAL DISK:

In the fIrst quarter of 1987, several majorarchival projects employing optical disk movedoff the drawing boards. The attention of thetraditional archives community was attractedby a nearly $1 million contract awarded by theNational Archives to SOC for an Wtical systemto store over one million military servicerecords. Although the. award culminates anumber of years of ~tudy, the ArchivalResearch & Evaluation Staff still considers it apilot project In February 1984, that staff fIrstrecommended digitizing pension, bounty landand military service r:ecords (over 8% of NARAholdings) and has been assessing thetechnology since.

Meanwhile, two larger projects have boldlycommitted to operational optical disk systems.The University of Syracuse has been awarded a

Foundation to convert 650 linear feet of adulteducation materials to an optical resource. Thearchive includes records of most of theassociations which have been active in theadult education field for the past 50 years, plusaudio and video tapes, photographs and thepapers of prominent academics in the field.Project Director Roger Hiemstra reports in theproject newsletter that QUALOG, a logicprogramming interface developed at Syracuseand now in use to assist in qualitative analysisof text materials, will pennit users to "ask aquestion and get a new response, ie. one thatwas not specifically programmed into thecomputer."

On an even larger scale, the University ofPittsburgh is proceeding with the digitizationand optical storage of full text data bases, imagebases and clinical data on all patients in itshospitals. The system will store all the datacollected during the course of patient stays on500,000 patients, or approximately 10 millionrecords and images. Using partitioneddata/image sets driven by microprocessorsattached to optical drives, the Vice Presidentfor Biomedical Infonnatics, John K. Vries,anticipates a system equivalent in power to asupercomputer for a cost of several hundredthousand dollars. The Medical Archives System(MARS) employs full-text automatic indexingwith artificial intelligenCe driven, thesaurusassisted indexing to capture medical dictation,directly word proCessed records and scanneddigitized text A retrieval system employingBoolean operatOrs and artificial mtelligenceparsers to express user queries in the samecanonical fonns stored dUring indexing,assures that users can use natural languageand get responses which are highly specific.Most of the system is now implemented, andhaving seen it, I am certain that we are lookingat the next generation of archives;

While numerous integration issues andvirtually all the stilndatds questions remain tobe resolved, the rate at which the technology isbeing adopted for official records keepingvirtually assures that the legal issues relatingto optical records will be resolved (by the bestavailable evidence rule?) in favor of the newtechnology. The IRS is, after all, keeping taxreturns in this fonnat. In Horida, theDepartment of CorporationS is implementing anoptical disk for storing fIlings of Horidabusinesses. New South Wales, has just giventhe go ahead to a $5.8 million registry of births,deaths, and marriages which will transfer 30million records to an optical disk system.

4 Archival Infonnatics Newsletter voLl, #1

MACHINE-READABLE: VIEWSby Thomas Elton Brown, PhD.

[ Editors Note: Tom Brown, Archivist,Administrative Staff, Office of theNational Archives has agreed to be aregular contributor to the ArchivalInformatics Newsletter. His column willfeature issues which he feels areimportant to archivists concerned withmachine-readable records. I amdelighted to have Tom reporting on afield he knows so well, and taught mostof us.]

KENTUCKY MACHINE-READABLERECORDS PROJECT:

A draft of a report on the Kentucky Machine­Readable Records Project (KMRRP) hasrecently been completed. The project was aneffort by Kentucky's Department of Librariesand Archives to gain archival control over thecurrent automated information systems in thestate government The excellent report,written by Margaret O. Adams (who has sincejoined the Machine-Readable Records staff atNARA) is titled Managing State GovernmentComputerized Public Records.

As archivists have attempted to cope with thegrowth and increasing technical complexity ofcurrent automated records systems, they havemoved away from inventorying computer filesas first proposed by Everett O. Alldredge in1971. In its place, three approaches have beendeveloped. These are: 1) microcontrol whichmanages automated systems at the data elementlevel; 2) system component level description tocontrol the different record componentswithin the automated systems; and 3) a capsuleapproach which develops inventories ofabbreviated descriptions of automated systems.

When the Kentucky Department of Librariesand Archives proposed to develop a datadictionary to manage the state's computerizedrecords, it was pursuing a strategy ofmicrocontro!. The initial concept of the datadictionary was an automated system whichwould track the data elements in Kentucky'sautomated systems. As this report documents,the project has moved away from the originalconcept of the data dictionary. While nothingin the report invalidates the microcontrolapproach, the report candidly acknowledgesthe practical problems which Kentucky facedduring the project and which led them awayfrom the original concept of the data

Ctlonary. lOut a common un erstan g

of what was the core of the project, it is notsurprising that communication was hampered.Another recurring theme concerned whowould provide information for input into thedata dictionary.

The report concludes with tworecommendations. Frrst, the author argues thatrecords officers should become "informationmanagers" who would manage the informationin an agency and a resource of that agency.Part of their responsibility would be to promotethe use of the agency's information by otherstate agencies and, if appropriate, release it tothe public. Secondly, the report urges that theKentucky Department of Libraries andArchives establish a Kentucky StateGovernment Repository and Service Center forMachine-Readable Records. This would be arecords center for computer files in whichproviding reference service to other agenciesand to the public would be stressed. It wouldprovide an opportunity for the archivists togain experience in processing, describing,preserving and referencing automatedmaterials.

Clearly, these recommendations raise twocritical points for the archival profession indealing with automated records. First,archivists and their colleagues in recordsmanagement need to become more concernedabout controlling the information in therecords rather than controlling the physicalmedium of the record. Secondly, most archivalinstitutions have not addressed the problem ofestablishing records centers storage andprocedures for computerized information. But,as the problem of non-current records whichare not eligible for disposition grows, varioussolutions will be proposed. The report on theKentucky project offers one of the firstconcrete proposals.

While raising two important issues, these tworecommendations reflect an underlying themeto this report. Namely the driving forcebehind an archival program for automatedinformation systems should be referenceservice. This stress on reference service can attimes give rise to troubling implications. Forexample, the release of some information maybe prohibited because of privacyconsiderations or other legal restrictions. Inthis situation, emphasizing reference mayresult in archival control missing suchrestricted information. Or a high degree ofcurrent research interest may be transitory,and the information which is the object of suchinterest may have little or no long term value.

Spring, 1987 Copyright by Archives & Museum Informatics 5

Conversely, information with significant longterm value may have little or no currentinterest. The emphasis on reference causes theresources to be devoted to the information withhigh current interest regardless of long termvalue. Such an allocation of resources may beinappropriate for' an archival institution.

FEDERAL GOVERNMENT ISSUESREPORT:

Archivists responsible for the administrationof computerized records frequently findvaluable information in obscure and unusualplaces. A recent Federal report is a case inpoint. The Office of Management and Budget,the General Services Administration, and theDepartment of Commerce jointly produced AFive-Year Plan for Meeting the Automatic DataProcessing and Telecommunications Needs ofthe Federal Government: Volume 1 (September1986). Despite the unlikely title, the reportcontains two discussions very relevant to anyarchivist concerned with automatedinformation.

The first item of interest is the report'sdiscussion of the current state of technologyand its forecast of future developments. Thistechnology overview appears in detaileddiscussions of several information systemsprojects and in a brief review of agencyinformation technology initiatives which laythe basis for a section entitled, "Forecast ofTechnological Trends". The report laterpresents an extremely lucid account of fourthgeneration languages and data dictionaries.Clearly, the applicability of such information

is far wider than just the Federalestablishment.

The second discussion of importance toarchivists is a section entitled, "InformationManagement: The Impact of New Technologies."The report argues that the creation,maintenance, and preservation of writtendocumentation is critical for operational andsocial needs. Claiming that informationmanagement must ensure the existence andavailability of this information, it outlines themanagement principles through theinformation life-cycle: creation/collection,maintenance, retrieval, and disposition ordissemination. These principles are thendiscussed in terms of the issues raised by threenew technologies: word processing, electronicmail, and electronic collection anddissemination of information. Thus the reportsuccinctly analyzes how automated processing

of full text information impacts the life cycle ofrecords.

DATA FILE DESCRIPTIONS ENTERINGNATIONAL DATA BASE:

The Inter-University Consortium for Politicaland Social Research (ICPSR) at the Universityof Michigan and the Research Libraries GroupInc. (RLG) have agreed to include thedescription of approximately 1,300 archivalcollections of machine-readable data mes inthe Research Libraries Information Network(RUN). ICPSR manages what is probably thelargest collection of machine-readable socialdata in archival custody. Placing thesedescriptions in a national bibliographicnetwork will make scholarly researchersaware of the existence of ICPSR's numeric datafiles and of the importance of computerizedinformation in general.

The project will begin with a machine­readable version of ICPSR's repository Guide,currently maintained in a local SPIRES datamanagement system. ICPSR will output aSPIRES file with descriptive information on thedata collections, and RLG will convert theSPIRES record to a MARC form. The MARCFormat will be the one specified for machinereadable data mes - MARC -MRDF. TheUniversity of Michigan Library will then addappropriate MARC-MRDF field tags not providedin the conversion, add Library of CongressSubject Headings, and assign other entries inaccordance with the Anglo-AmericanCataloging Rules. This retrospective conversionis projected to be complete in August. As ICPSRaccessions new mes, descriptions of them willbe added to RLIN.

STORAGE TECHNOLOGY & REFERENCESERVICE:

Archivists at traditional machine-readabledata repositories have fairly well establishedappraisal criteria, processing guidelines,descriptive standards and preservationprocedures. However, most archives are stillusing reference procedures established in thelate 1960's by data archives and data libraries.The current standard seems to be to provide acopy of the data me in a fixed length, flatformat on magnetic tape encoded to theresearcher's specifications. As technology hasprogressed, data libraries and data distributorshave developed other options which archivistshave not adopted. For example, the Census

6 Archival Infonnatics Newsletter vol.l,#l

Bureau and the National Technical InfonnationService provide selected infonnation of floppydiskettes. Others provide interactive, on-line·access to selected numeric infonnation.Reference strategies for computer files.especially full-text infonnation, will hopefullybecome a major research area at traditionalrepositories as they examine the possibility ofexpanding their reference service options.Presumably, an initial step would be to examineoptions which data libraries and data archivesare presently offering.

In this connection, the Data Archives onAdolescent Pregnancy and PregnancyPrevention (DAAPPP) has announced a newservice using the Bernoulli Box. This is a small,portable device that connects to amicrocomputer. It reads and writesinfonnation on magnetic cartridges abouttwice the size of a floppy diskette. Physicalsize, however, is deceiving. Where an ffiM-PCfloppy diskette can hold up to 360,000 bytes ofinfonnation and an ffiM-AT floppy can hold upto 1.2 million bytes, a single Bernoulli Boxcartridge can hold up to 20 MB. The discountprice of a Bernoulli box with two 20MB drives isapproximately $2,600. Once the initialinvestment is made, removable 20 MBcartridges can be obtained for only $65 each.DAAPPP is now offering all of the dataappropriate for microprocessors on three 20MBBernoulli cartridges for $750, or 40% less thanthe floppy diskette versions. Included are themachine-readable raw data and SPSS/PCprogram statements with printeddocumentation.

Clearly the development of low-cost, large­volume storage media is changing the optionsarchivists have in providing computerizedinformation to researchers.

Thomas E. Brown

ART TASK FORCE· LEARNING OBJECTIVES:The SAA Automated Records and Techniques Task

Force has updated its "Learning Objectives and CourseContent" Guidelines for courses on Basic ComputerConcepts, Automated Techniques and MachineReadable Records. As of October 1986, "archivistshave used this curriculum for the basis of 32workshops during annual SAA meetings." Inaddition, the SAA Task Force on Automated Recordsand Techniques has used the curriculum to present atwo day workshop on machine-readable records onfour occasions and a two day workshop on automatedtechniques, twice. The Guidelines are reprinted herewithout part A, the computer literacy course.

An Educational Program toTrain Archivists inAutomation:

B. Automated Techniques1. The archivist will understand how an automated

system can organize and manipulate information.2. The archivist will learn how to determine if the

implementation of an automated retrieval system usinginfOImation in manual systems will produce desiredoutputs.

3. The archivist will understand the principles of thedecision making process regarding the acquisition ofan automated information system.

4. The archivist will create an exercise database on amicrocomputer using commercial databasemanagement software.

5. The archivist will outline procedures to evaluate,maintain, and expand an operational database.

6. The archivist will understand how to integrateinformation in an automated information system andthe subsequent impact on providing reference serviceto archival materials.

C. Machine Readable Records1. The archivist wilileam to inventory the

components of and then describe automatedinformation systems.

2. The archivist will determine the informational andevidential value of a machine-readable file through ananalysis of a) units of analysis, b) the level ofaggregation, c) the difference between administrativeand survey data, d) linkage potential and, e) therelationship of similar information in hard copy format.

3. The archivist will learn to develop records controlschedules for automated record systems based uponinformation gathered during the inventory and thedecisions made during the appraisal.

4. The archivist will detennine if sufficientdocumentation exists to accession a file into archivalcustody.

5. The archivist will determine how to processmachine-readable records in order to make themavailable for research.

6. The archivist will understand the types ofinformation needed to describe machine-readable datafiles and how to use standardized formats in describingmachine-readable data files.

7. The archivist will discuss the dissemination offIles with restricted information.

8. The archivist will discuss the researchcommunities for machine-readable data.

9. The archivist will discuss preservation techniquesto ensure the integrity of machine-readable files inarchival custody.

10. The archivist will discuss trends in computer andinformation systems technology.

Spring, 1987 Copyright by Archives & Museum Informatics 7

WHAT AREIIS INFORMATICS?And especially, what/who is Archives &Museum Informatics?

In November 1986, I fonned Archives &Museum Infonnatics as a sole proprietorship,which, as the State of Pennsylvania so aptlyputs it, means that, at the moment, Archives &Museum Infonnatics is a "fictitious name" forDavid Beannan. Perhaps, someday, others willbe involved as well.

Archives & Museum Infonnatics was fonnedto conduct research and publish. I adopted thetenn "Infonnatics" from an emerging usage inbiomedicine where the importance ofinfonnation technologies (like computers),infonnation techniques (like full-text retrievalor digitizing radiographic images) andinformation theories, especially those oflinguistic analysis, artificial intelligence,indexing and retrieval, are coming together innew ways of practicing medicine. The tenn,while relatively new, is used enough so that inthis issue of the Newsletter, without my makingany effort to do so, it appears in the title of one(other) journal and the title of a person's job.To me the importance of the concept is that isreplaces automation, or records, orcomputerization with a system oriented view ofthe synergism of infonnation based activities.

Archives & Museum Informatics is also aconsulting firm, and the tenn "Infonnatics" inthe name expresses a philosophy of looking atthe cultural information missions of archivesand museums broadly in order to addreSs arange of new approaches, techniques, andtechnologies which can enhance anorganizations prof11e and achieve its mission.

In addition to publishing and consulting, Ihave developed courses to teach a number ofskills which I believe are overlooked by collegeprograms which are available to professionalsin their own communities and by thespecialized training programs thus far offeredby SAA, AASLH, ARMA, NAGARA or the museumassociations. These skills range from Planningfor New Technologies (a course for managerswhich focusses on risk assessment, needsanalysis, standards, introducing change inorganizations etc.), to UnderstandingVocabulary Control (a course for archivistsinvolved in record description and/orreference, which focusses on the kinds ofvocabularies which can be controlled andeffective tactics for controlling them.) Acourse on Planning for Automation whichpresents each step in an automation project

and gives hands on experience in fonnsanalysis and writing of functionalrequirements is designed to help fonn a teamin an archive in which a decision "to automate"has been made bUlthe implications of thedecision have yet to be worked out

Originally I had intended to give all thesecourses as public seminar/ workshops, but thecosts to participants of travelling to Pittsburghhas convinced me to offer them instead understate archives, regional organizations, oruniversities sponsorship where I have totravel, but many of the participants are nearhome. I welcome inquiries about these andother seminars.

Over the past three months, Archives &Museum Infonnatics has been involved in anumber of projects for non-profitorganizations which I will report in a futureNewsletter, including:- Mapping a proposed data dictionary fordescription of Architectural Drawings into aMARC fonnat for possible use by the I.PaulGetty Art History Information Program.- Developing Policies & Procedures for theEstablishment of a Software Archives by theComputer Museum (Boston) in conjunctionwith the Smithsonian Institution. This is phasetwo of a project, the fllSt phase of which was afeasibility study submitted to the sponsors inFebruary.- Designing a method, for the National SecurityArchive, of indexing at an item level verylarge numbers (30,000+ p.mo.) of recentlydeclassified documents relating to AmericanForeign Policy.-Assisting the Florida State Archives staff todevelop an automation plan, by conducting aworkshop with them on automation planning.- Critiquing and analyzing tenninologyproposed for inclusion in the Art andArchitecture Thesaurus.

Although Archives & Museum Infonnatics isclearly a commercial proposition, I hope that itcan also make a contribution to archives andmuseum automation through conducting kindsof research activities which cannot besupported by cultural institutions acting ontheir own. One aspect of this effort is toencourage contributions by others to thisnewsletter in order to increase the speed offlow of technical infonnationwithin ourcommunity. Another aspect is to ask for yourideas for future research and publications, andto provide means to pool resources in order toconduct such studies. I look forward to hearingfrom you.

8 Archival Infonnatics Newsletter vol.1, #1

THE ARCIllVAL MANAGEMENT OF MACIDNE-READABLERECORDS FROM DATABASE MANAGEMENT SYSTEMS:

A TECHNICAL LEAFLET

By Thomas Elton Brown & William A. ReaderSociety of American Archivists, Task Force on Automated Records & Techniques

lI,

INTRODUCTION:Data Base Management Systems (DBMS's) pose

problems for records managers and archivists. TIllstechnical information leaflet addresses theseproblems. It begins with a discussion of archival

issues associated with DBMS's. Since most of these

concern the transfer of DBMS- managed information to

an archives, another section recommends transfer

procedures for specific DBMS packages based on a

survey of vendors of DBMS packages. Vendors were

asked to recommend a means by which information in

one of their systems could be transferred to an

archives. These procedures are presented here so thatan archivist or records manager can convey these

instructions to the database administrator of a system

which contains historically valuable information.

Using the recommended procedures, the database

administrator will then be able to overcome the

technical difficulties associated with transfer of

archivally valuable DBMS information to a repository.

Since the universe of DBMS software packages is

constantly changing, the leaflet also amlOunceS a

clearing house function. The purpose of the clearing

house is to maintain the currency of the leaflet

information by providing a point of contact to correct,

refine and supplement the recommended procedures

for transferring information from specific DBMS

packages to an archives.

I, Archival Problems Created byData Base Management Systems:

In recent years, organizations generating archival

data have created databases and installed database

management systems at a phenomenal rate. Unlike

conventional archival tape files which are software

independent and can be processed on any computer

system, DBMS-managed databases organize their datain ways that confer benefits for users but create

serious potential archival problems.

There are several advantages to placing data in a

database. One does not have to access the entire

database in order to use just some of the data. Thus a

bank posting check transactions needs only to access

checking accounts and does not have to read the other

files in the database. TIlls cuts down on processing

time, permits simultaneous processing of multiple

applications and can limit access to sensitive data

areas of the database to only certain users. Without a

DBMS, applications programmers must know thelocation and structure of the data when writing a

program to access or process it. With a DBMS,

programmers need only know the names of the data

elements and thus can focus on turning outapplications programs. All this, in t1ml, cuts costs

and improves operating efficiency.

In a DBMS-managed database, data elements and

records are physically delimited by control

characters embedded in the data. Records consist of

either physically separated data fields or data

elements in different tables which are linked into a

logical record by pointers or database keys. Records,

in turn, are logically organized into sets which,

according to the type of DBMS, consist either ofidentical record types defined by database keys,

pointers which locate the record's constituent

elements, or hierarchically linked record defined by

pointers tying one record to another.

These DBMS characteristics, however, have practical

implications which cause potential archival problems.

First, in a DBMS-managed system, there is nocorrespondence between the physical and logical order

of the data. Such relationships must be created either

by the DBMS software, which maintains defmitions ofdata elements, records, sets and linkages, or by the

routine that dumps the data into an archival file.

Second, different generic types of DBMS's give very

different definitions of data elements and records,

provide different means of linking data and records

together to provide data to the user, and supply

different delimiters and pointers to indicate the

specific physical location of data fields, records and

table elements. In addition. even within one class,

software-defined data fields, record delimiters, and

keys may differ from specific system to specific

system. Consequently, data managed by one DBMS

carmot be processed by a computer using a differentDBMS.

Recently, the problem of DBMS data transportability

Spring, 1987 Copyright by Archives & Museum Informatics 9

has been aggravated by computer and DBMS

manufacturers embedding certain recurring softwareoperations on special purpose computer chips to free

up computer memory for other instructions and data.

As more software becomes embedded in computer

hardware, the data stored in DBMS systems becomes

hardware specific.

Thus, for an archival institution concerned with the

permanent preservation of historical data, thearchiving of database information presents a

substantial problem. Specifically, how does one

transfer historically valuable information resident on

a specified DBMS to an archives? Traditionally,

archives have sought to avoid becoming repositories of

obsolete hardware and software systems because they

have neither the funds, expertise nor space to acquire

and maintain antiquated computers and programs.

Consequently, archives have acquired machine­

readable data in flat fixed-record length formats free

of any extraneous control characters or delimiters.

To follow this practice for DBMS data. archivists

must determine whether a DBMS has an "unload"

routine that will output all or selected fields into

linear sequential fIles without any embedded control

characters. Unfortunately, many DBMS's have "unload"

routines which correlate physical and logical order,

but leave extraneous control characters which can beread only by the DBMS that Unloaded the database.

Thus, their output is only transportable to another

computer using the same DBMS. To make the unloaded

data transportable, one .has to write a program to

reinterpret the control characters, and, consequently

output the information in a software independent

format This progr~gwould be tedious, time

consuming and expensive.

Even if the creation of an archival. linear,

sequential; and fixed-length record is possible, there

is often a related question of practicality. Since

DBMS's are structured to minimize duplication of data

elements, the creation of linear sequential files from a

DBMS will result in numerous separate files or a

massive linear sequential file in which data fields are

constantly repeated. For example, a DBMS containinga sample survey of cohort data would have separate

fIles of geographic area data. household data. persondata, and separate time period interview data for each

person. Creating linear sequeI!tial files couldnecessitate separate sets of geographic, household,

person and interview fIles. Or it could mean thecreating of an interview fIle incorporating the data

elements from all the fIles. In this case, geographic,

household, and person data common to each separate

interview would be repeated for each interview record.

This would result in a large quantity of redundant

data.

The extent of the archival problem with a DBMS is

dependent on the paricular software package that

manages the information. To examine the issue in

terms of specific DBMS's, the Society of American

Archivists Task Force on Automated Records and

Techniques surveyed vendors of Data Base Management

Systems. The goal of the survey was to determine what

procedures the vendors would recommend for

transferring DBMS-managed information to an archivalrepository.

ll. Survey of Vendors:To determine which DBMS vendors to survey, we

reviewed Datapro Research Corporation's In-Depth

Analysis of Leadin~ Database. Data Mana~ement and .Data Dictionary Systems (1985) to identify vendors

with a large number of installed systems. Letters were

sent to those vendors in June 1986, asking how

information being managed by their DBMS product

could be transferred to an archives. The letter invitedvendors to propose any or all of three possible

strategies: output of a flat, rectangular format free of

extraneous control characters; use of an international

standard designed for exchange of information

between systems (1S0-8211); or any other means the

vendor would recommend.

Initial responses varied greatly in quality. As a

result, follow up letters or telephone calls were used

to clarify the initial responses. Tentative results

were presented at the Society of American Archivists

annual meeting in September 1986 and follow-upcontacts continued throughout the fall.

It is noteworthy that no vendor reported

implementing IS0-8211, the International standard

for the exchange of data fIles. Among those vendors

who recommended any specific procedure for

transferring information from their system to

archives, almost all focussed on techniques for

creating a linear, fixed length data fIle on magnetic

tape.

Nothing in this report should be construed as anendorsement of any particular systems by the authors,

the SAA or the publisher. Furthermore, the exclusion

of any particular system does not mean the vendor

does not have a large number of systems in use;

vendors with numerous installations may not have

responded to the initial letter and a lack of time may

have prevented follow-up. The instruction for

transferring information from a specific DBMS to an

archives may need correction. The recommendedprocedures are based on discussions with the vendors'

representatives, review of user manuals or

interpretations of letters, any of which can be

misinterpreted. In addition, no tests of the vendors

recommended procedures have been made to

independently validate their claims. Because of the

potential for misinterpretation, because no tests have

been nul, and because vendors are continually adding

to and altering their products, these procedures will

10 Archival Infonnatics Newsletter vaLl, #1

~~~~~~~~---~~----------------_.--

l,--"'

IIr",

need updating. Also, this technical infonnation leaflet

does not address all of the DBMS's which archivists

and records managers will confront and will need to be

updated as more DBMS's are developed and

encountered and as archivists and records managers

validate these procedures in their experience with

different systems. To- correct or expand on this

report. archivists and records managers are invited to

contribute to a "Clearinghouse" by contacting the

authors, clo ARCHIVES & MUSEUM INFORMATICS,

5600 Northu:mberlandSt, Pittsburgh,PA 15217.

m. Vendor Recommendations:

APPLIED DATA RESEARCH, Inc.

ADRIDATACOMIDB

This DBMS package provides a single utility, called,

DBllTI.-TY, for creating and maintaining Ihe database.

It includes two routines which have applicability for

creating output capable of being transferred to

archival custody.

The first routine is BACKUP. While the primarypurpose of this routine is to create security back

copies of Ihe database and its index. careful selection

of options relating to DATACOM's "keywords" can

produce a variable length, sequential file which

contains standard mM four-character-Iength

identifiers. Specifically, the DBUfLTY program should

include Ihe following command formats: CMPRS=NO

and SEQ=NATIVE. Since this routine can INCLUDE 01'

EXCLUDE specific tables or files from the database.,

the execution of one routine could output all of me

infonnation of archival value from the database.

The second routine is EXTRACT. This routine writeS

each table as a fixed-length record in fixed-block

fonnat. Extract Ihe table or file in a native key _

sequence (SEQ=NATIVE). Since only One table or me

can be extracted by this routine, a separate execution

is required for each table with infonnation of archival

value.

CULLINET SOFTWARE Inc.

IDMSIR & Culprit

Culprit is the report generator associated wilh

IDMS/R, the Integrated Database Management

System/RelationaI. Approximately 95% of installed

IDMS/R packages have Culprit Through Ihe use of

Culprit, a portion of the database is specified, then a

specific database, Ihen the data, and then the data to

be written to magnetic tape. This is known as

SELECTIVE COPY. The data will be written to tape in a

decompressed format and will be accompanied by a

"listing" of the copied data elements. Since the data is

decompressed, this listing serves as a record layout

This process requires the selection of each data

element for transfer.

Infodata Systems Inc.

INQUIRE

This DBMS is primarily for text rather than numeric

information. It can unload data files into a sequential

file fonnat through the SAVE command. It selects

particular data fields from the database for output,

beginning in a specific field location and in-a

linearized format If outputting textinfonnation, the

return characters could be interpreted to output the

infonnation in paragraph fonnat or could be

suppressed to output the infonnation in a long

sequential string. _This procedure requires the

selection of individual data elements. Another

possibility is to dump the infonnation into Wodata

Standard Input (lSI) format which is a field delimited

me in which field names appear in columns one

through eight of an eighty column card image and the

value of each field wraps between columns ten and

seventy-two. To employ this utility, execute SAVE ISI.The system prompts for the database name, Ihen asks

for record description to select certain records from

the database, and then unloadS that database to a

sequential file in lSI format Since Ihe me is first put

out to disk, one would have to use the operating system

utilities to move the file to tape for archival storage.

Information Dimensions Inc.

(Formerly Battelle Memorial institute)

BASIS

This DBMS is primarily for text rather than numeric

infonnation. The package includes at least ten optional ­

modules. One of the more commonly installed modules

is REPORT. Within the module, REPORT WRITER

specifies each field and the field length. In the

execution, the data definition is automatically

provided for a complete codebook. If one is dealing _

with only numeric data, the REPORT module can define

an output me in the fonnat called for by a statistical

analysis package such as SPSS or SAS. In this

situation, the BASIS data could be moved as an SPSS or

SAS file. Then the utilities within Ihose statistical

packages could produce a sequential, fixed-length,

file.

I.B.M. Corporation

IMS (Information Management System)

IMS is one of the oldest [)BMS's; Ihe first installation

was made in March 1971. Each IMS system has a LOAD

or UNLOAD utility for use in main~g and

compressing the data in Ihe DBMS. However, using

UNLOAD will produce a sequential file embedded wilh

IMS control characters and thus still dependent on the

IMS software. Therefore, IMS does not nonnally have a

Spring, 1987 Copyright by Archives & Museum Infonnatics 11

utility which can create a rectangular output file free

of control characters. Since 1984. IBM has beenmarketting a relational database system called DB2.

In order to permit IMS users to migrate to DB2, a

DATA EXTRACT utility was created for IMS to produce

a VSAM format of IMS data. The use of the DATA

EXTRACT utility allows a choice of transferring acomplete table or just portions of a table or file. The

DATA EXTRACT utility requires the user to specifythe path through the data hierarchy. Since a VSAM

format can be moved by mM system utilities to a

rectangular tape format, the use of this utilityprovides the intermediate step for the creation of a

rectangular data file. However. very few IMS

installations will have acquired this utility. since its

purpose is migration to DB2

LB.M.CorporationDB2

Since the DATA EXTRACT utility is the DB2 approachto maintaining and compressing information, almost

all DB2 installations will have it. This utility willoutput from the DB2 software a VSAM format of DB2

information. The use of DATA EXTRACf allows a

choice of transferring a complete table or just portionsof a table or file. The DATA EXTRACT utility allows

the user to relate different files in differentcombinations. Since a VSAM format can be moved

through mM system utilities to a rectangular tape

format, DATA EXTRACT provides the intermediate

step for the creation of a rectangular file.

Oracle CorporationOracle

EXPORT is an Oracle database file backup facility

which produces an encrypted. compressed file. Users

may send flat sequential datafiles from SQL*PLUS onto

tapes. Oracle also contains a standard utility calledORACLE DATA LOADER (ODL) to read and store flat

sequential ASCII or EBCDIC files. but apparently doesnot have a routine for the reverse process.

SAS Institute Inc.SAS (Statistical Analysis System)

Originally designed for statistical research for

academics, this package is increasingly used forgeneral data management in other organizational

settings. The package consists of collections of

instructions. called steps, and of individualinstructions within steps called statements. To create

a rectangular fIle (or "raw data fIle" in SAS

terminology) from a datafIle in the SAS system (or a

"SAS dataset" in SAS terminology). one would use the

DATA step. With the DATA step, two statements are

necessary. The first, is FILE, a statement which will

specify the format of the output fIle. A FILE

statement can have different options. Whether and

how these options are used will depend on how the

information in the SAS dataset is structured, e.g. a

hierarchical file. The second statement is PUT. The

options with the PUT statement can specify whetherthe entire file or just selected variables within a file

will be moved to the output file. The vendor promotesSAS as self-documenting. As a result, a record layout

and codebook can be created using the CONTENfS

statement in the PROC step.

SAS InstItute Inc.System 2000 DBMS

This DBMS does not have a utility program through

which a sequential rectangular file can be created.

After SAS Institute Inc. purchased System 2000, about

2 years ago. a procedure was developed by which a

System 2000 hierarchy can be incorporated into a SASdataset. The routine is PROC S2K. When using thisprocedure, it is necessary to specify the path or pathsthrough the hierarchy. With the creation of a SAS

datset, one can use SAS procedures to produce alinearized, sequential file.

Software AG of North America, Inc.ADABAS

This DBMS does not have a routine to use in the

production of a fixed-length output file. However.

over 90% of ADABAS installations have a programming

language called NATURAL. A Three statement program

in NATURAL would produce an output file. A READ

statement would specify the file or tables; a KEYstatement would identify the variables being selected;

and a WRITE statement would stipulate that the outputis to be a fixed-length record. The KEY command must

be followed by the database names of the variableswhich will be transferred to output, thus the system

permits only the selection of individual variables

rather than complete tables or files. In the 10% ofADABAS facilities which do not have NATURAL, a

COBOL or PLl program can be written to effect the

same output.

UNISYSDMS-n

This DBMS, a product of the former Burroughs

Corporation, includes ERGO (Extended Retrierval with

Graphic Output) which is a report generator. Within

ERGO, the EXTRACT commands could produce a flat,

sequential, fIle on tape. But to exercise this program,

it is necessary to identify each data element to be

transferred to the output file.

12 Archival Infonnatics Newsletter voU, #1

CONFERENCES:

The Quarterly Update will not pUblish acomprehensive calendar of upcomingconferences; instead it willllst conferenceswhich we intend to attend and may report onin future issues or conferences reported tous as having sessions of special interest toreaders.

May 7-9. 1987 Mid-West Regional ArchivesConference in Chicago. Linda Evans reports that themeeting will include exhibits of LS2000,MicroMarc/AMC, ZyIndex. and Cactus Software.

May 8-9. 1987 Mid-Allantic Regional ArchivesConference in Baltimore. Sessions on Automation forReference at NARA (red Weir) and NY University (romFrusciano); on expert systems in archives (WilliamHolmes. Pat Molliolt, David Bearman); and on PC's andMARC/AMC (Donna Wells & David Anderson).

May 17-20. 1987 American Society for InformationScience Mid-Year Meeting. Kings Island. Ohio. Focuson laser optical disks and video information systems

May 19-22, 1987 International Association of SocialScience Information Service and Teclmology (1­ASSIST); 13th annual conference. Vancouver, BC;Sessions include workshop on accessioningcomputerized information into traditional archives.Tom Brown plans a report in the next issue.

June 4-6, 1987 Association of Canadian Archivists,Hamilton Ontario.

June 7-11, 1987 American Musemn Assoc. Meeting,San Francisco.

July 11-13, 1987 International Conference onDatabases in the Hmnanities and Social Sciences,Montgomery Alabama, includes keynote by FrankBurke on ISO standards 8211 and electronic archivesand 6 other papers on electronic archives projects.

July 22-25. National Association of Gov. Archivists &Records Administrators. Atlanta Georgia. Pre­conference workshop on "Starting an ElectronicRecords Program" with Margaret Hedstrom, postconference workshop on "Document Base ManagementSystems" with Edward Johnson, and numerous sessionson automation in between, including one devoted todiscussing the RLG/Seven State Archives Project.

ART LffiRARIES AND ARCHIVALAUTOMATION DEVELOPMENTS

The meeting of the Art Library Association ofNorth America (ARLIS/NA) in OC in February,included demonstrations of a large number ofcomputer projects which indicated the impactwhich emerging standards are having invisual arts cataloging. The revision of theMARC-VM format, the publication of Betsy BetzParker's: Graphic Materials:Ru1es forDescribing Original items in HistoricalCollections, and the growing acceptance of theArt & Architecture Thesaurus were revealed.along with the increasing sophistication ofsystems employing these standards.

Cathy Whitehead of the Art & ArchitectureThesaurus and Amy Lucker of the BrooklynMuseum demonstrated a generic non-bookformat, mapped to MARC tags, using MARCONsoftware by AIRS Inc. The system providedactive authority control from seven AAThierarchies to the appropriate MARC field,supporting both alphabetical and syndeticdisplays of AAT terms. Jeanne Keefe Watkinsonreported on the Rensselaer PolytechnicInstitute SPIRES based system which employsthe AAT Styles and Periods hierarchy to createcall numbers. Merrill Smith demonstrated adatabase, using Informix, to the MI.T. RotehVisual collection videodisc which is alsodesigned to employ AAT descriptors. A linkageof AAT, MARC and videodisc was demonstratedby Angela Giral of the Avery Fine Arts Library(Columbia University). MARC based records ofoutdoor scu1pture were demonstrated on amodified GEAC library system by ChristineHennessey of the National Museum ofAmerican Art. The system is designed to employAAT terminology in on-line authority controlwhen that database can be mounted.

Numerous other projects of note weredisplayed, including a database developed byPat Barnett of the Metropolitan Museum of Artwhich is a "Clearinghouse on Automation inthe Visual Arts" maintained on MARCON.Several slide cartaloging programs, using

Smart Software at SUNY Cortland, SPIRES atRensselear Polytechnic Institute,· and Data Easeat the University of Oregon, were featured. The

Computer Project Demonstrations testify to avery active Special Interest Group withinARLIS with which print archivists have muchto share.

Spring, 1987 Copyright by Archives & Museum Infonnatics 13

Archival Informatics Newsletter

FOLLOW UP ON THE NEWS:

VALENTINES DAY ERASURE:

The New York Times of February 15, 1987reported that all the computerized records ofthe town of Prescott Valley, Arizona had beenerased "leaving officials with no idea howmuch money has been spent this year or howmuch cash the town has left." I wondered - doany State Archives offer a security magnetictape storage program to local governmentsanalogous to their security microfl1mprograms?

DISCLOSED ERASURES:

The lesson is still unlearned, even if AnnGorsuch was burned by electronic tracksearlier in this Administration! Now we discoverthat Admiral Poindexter and Colonel Northforgot about PROFS, the office automationsystem through which their internalcorrespondence flowed. Much of what theTower Commission learned, came from thePROFS audit trail; Government archivists maywant to consider what this implies for thehistorical record, since such transactionalaudit trail records are nearly certain to haveextremely short retention periods, if they areeven sCheduled.

ERASE THIS ADVICE:

The most recent edition of the FellowesManufacturing Company's RecordsManagement Handbook. distributed as a freeservice, has been updated to includeinformation about the preservation ofmagnetic media Unfortunately they neglect tonote that magnetic tapes should be rewoundunder constant tension at least annually andchecked for errors, which, if found, call forcopying onto a new tape. When I pointed outthe problem, a Fellowes spokesman promised totake this into account "when it comes topublishing our tenth edition". Until then,beware or fmd your tapes erased before then.

14

ARCHIVES NEED NOT APPLY:

What I read about the Library of CongressCorrespondence Reference Program, whichnow refers reference questions received by LCto 41 State Libraries so that answers to patronqueries directed to LC may be provided by thepatron's local library service, sounded like aterrific idea. So I wrote to ask if they hadconsidered cooperating in a similar programwith state archives.

R. David Myers, Head of the Section of theGeneral Reading Rooms in charge of theprogram, answered that they had not, but thatarchives would be welcome if the task could becoordinated through Trudy Petersen at NARA.If nothing else, it would be an excellenteducation for LC staff; his answer ("nor do wereceive many questions related to archives")suggested that the idea that archives could bean information source, rather than the subjectof a query, was novel!

SOUNDS UNREAL? - MIGHT BE!

Optiram Ltd. is offering a text conversionservice to publishers and archivists whichthey claim can scan any typefont and evenhandwritten texts and convert themautomatically into ASCII characters withlimited up front "teaching" of the font orhandwriting style. Considering the dimensionsof the breakthrough this implies, one wouldexpect Optiram to be aggressively licensing itstechnology. Instead it is acting only as aservice bureau, and won't demonstrate itssystem or answer my mail about its technology.Personally, I think we are dealing withanother company which uses Phillipine,Korean or other third world data entrypersonnel, at least until they prove otherwise,but meanwhile they are offering veryattractive prices for transcription of original,photocopied or telefacsimile texts - between$.95 and $1.75 per 1000 characters ($950 -$1750per megabyte or approximately 400 pages oftyped material). Discounts are available forlarger volumes. This and other similarservices, usually explicitly involving off-shoredata entry, have dramatically changed thefinancial realities of major retrospectiveconversion projects for archives - anyone upfor retrospective conversion of fmding aids?

YoU, #1

';

IN-BOX

This Department lists a selection of Itemsreceived by Archives & Museum Informaticsin recent months. It is not intended to be acomplete bibliography of archivalautomation.

REPORTS:

American Library Association (ALA WashingtonOffice, 110 Maryland Ave. NE, Washington, OC 200(2),

Less Access to Less Information By and About the U.S.

Government: 2 - A 1985-1986 Cbronolo~, 1987,33pp.A sequel to the 1981-1984 Chronology. A quite

frightening picture emerges from these pages.

Council on Library Resources (1785 MassachusettsAve., NW, Washington OC 20036), Scholarship in the

Electronic Age: A Selected Bibliography on Research &

Communication in the Humanities and Social Sciences.Compiled by Anita Lowry and Junko Stuveras,

February 1987, 54pp. Exceptionally interesting

selection with excellent annotation important to

archivists trying to understand the ways in which

automation is changing scholarly research and

communications.

Also, Costs of Microfilm Preservation at ResearchLibraries: A Study of Four Institutions, by Paul B.Kantor ($3 prepaid).

Dele (6565 Frantz Rd., Dublin OH, 43017-0702),

Campus of the Future, March 1987; proceedings of a

conference sponsored by DeLC and the Johnson

Foundation June 22-24, 1986. TIlls conference heard

a number of very important papers on the ways in

which information technologies are changing academia,

including one by Douglas Van Houweling (Vice Provost

for Information Technology at the University of

Michigan) entitled 'The Information Technology

Environment in Higher Education" which should be

must reading for university archivists.

National Research Council, Preservation ofHistorical Records. Washington OC 1986. Technical

discussion of media characteristics and current

research.

University of Waterloo, Institute for ComputerResearch, Microcomputer Database Management

Systems: Application to Cultural and Herita~e

Information Tasks. A Report Concernin~ Tests of 24

Commercial Software Packa~es. by E.M. Avedon et.aI.,

March 1986. The continuing value of this report lies

in the five test models, based on different types of

cultural management tasks, developed for the

evaluation and the discussion of features of products

which make them suited to doing specific tasks well.

($5 to: Canadian Conference on the Arts, 126 York St.,

Suite 400, Ottowa, ON, Canada, KIN-5T5).

NEWSLETTERS:Humanities Communication Newsletter - ISSN 0269­

0519 (Office of Humanities Communication, University

of Leicester, Leicester LEI-7RH, ENGLAND) edited by

Dr. May Katzen, is the most comprehensive single

source on European computerization projects in the

humanities. Since 1983 it has grown from a two page

flier to a 50 plus page bi-annual journal.

Museum Archivist (Kathleen Robinson, Editor,

Museum of Fine Arts, P.O.Box 6826, Houston TX

77265) is the new publication of the museum

roundtable of the Society of American Archivists. Its

first issue, December 1986, contained a variety of

useful reports.

Scholarly Communication (American Council ofLearned Societies, Office of Scholarly Communication,

1717 Massachusetts Avenue NW, Suite 401,

Washington OC 2(036) has reached issue seven and

continues to report in depth on a wide range of

scholarly publishing and library issues.

ARTICLES & BOOKS:

Bailey, Robert, Archives Automation 1975-1987, For

the Record ..Newsletter of the Dlinois State Archives,Vo1.9(2). Winter 1987 p.1-4. Describes over a decade

of automation projects at the Illinois State Archivesand includes a section in which the author complains

that too much emphasis is being placed on MARC AMC.

Baskerville, Peter A. and Gaffield, Chad M.,

Archives. Automation & Access: Proceedin~s of aconference at University of Victoria, British Columbia,

March 1 & 2, 1985, University of Victoria, no

pub.date., 152 p. This conference included papers on

automation issues in archives by Terry Cook, RichardJanke, David Bearman, Terry Eastwood, Hugo Stibbe,

Susan Rosenfeld Falb, Sue Gravel, W. Theodore Durr.

David Mattison, Michel Roberge and ihe editors. Most

of these articles remain timely and are not available

elsewhere.

Cook, Michael, The Mana~ementof Information FromArchjves. London, Gower Publishing Co., 1986, 203 p.

plus bibliography and index. Cook's intention is to

cover the entire life-cycle and all processes. Fromthis side of the Atlantic, his observation on how little

has been transmitted on archival automation ispowerfully reinforced; at least we can now better

understand the English side of the equation.

Spring, i987 Copynght by Archives & Museum Infonnatics 15

Archival Infonnatics Newsletter

DeSalvo, Daniel, An Interview on Local Area

Networks with Dr. Frank Burke, Acting Archivist ofthe United States, Telematics and Informatics, vo1.3(2)

1986, p.93-100. An informal, slightly ideosyncratic,

discussion. De Salvo's aricle on NARA expert systemsin the same journal, vol. 3 (1) p.25-38 is well worth

reading for the insight it provides into NARA

priorities.

Jasco, P., Szucs, A., and Varga, S.; Micro-CDS/ISIS: A

Bibliographic Information Management Software from

UNESCO, Microcomputers for Information Manaeemem,3(3), 173-198. Describes the microcomputer version

of ISIS, a public domain bibliographic system now

used in mnnerous archives (especially film archives)

around the world.

Kimball, Margaret, Workflow for Processing

Manuscripts in Automated Systems, Rare Books &Manuscript Librarianship 1(2) Fall 1986, p.117-126

is an edited version of a paper given at the 1985 SAA

meeting on Stanfords' early RUN MARC-AMC

experience.

Manns, Basil & Swara, Tamara, Books to Bits: Digital

Imaging at the Library of Congress, Journal ofInfonnatjon and Image Management. v.19 (10), 1986,

p.27-32 report that the Library of CongressPreservation Microfilm office is finding digital

scanning more cost effective than microfilm! Other

issues relating to the technology are still being

explored.

Sermis, Mark & Givner, Christine, Implications of

Analog-to-Digital Technology for Social Sciences: AReview & Preview, Social Science Microcomputer

Review, v.4 (3), Fall 1986, p.295-309. Yes, it will

change the ways we ask questions and the way we

answer them too. This and the Archeolo~cal

Computinll Newsletter, which fve recently begunreceiving, introduce us to a whole new world of social

science methodology.

Unsworth, Michael E., A Review of the National

. Inventory of Documentary Sources in the United

States, Microfonn Review. 15(4) Fall 1986 p.232-239

is the first extensive review of NlDS I know of. The

author is relatively positive overall but doesn't ask

how any national inventory which includes only 791 of

the 10,000+ collections at the Library of Congress or

196 out of more than 425 published NARA registers

can substitute for a comprehensive tool.

Weber, Lisa, Status Report on SAA's Automated

Information Program, SAA Newsletter, March 1987,

p.7. Reports on SAA workshops on MARC-AMC, the

clearinghouse on archival automation. and the results

16

of the January MARBI meeting which adopted changes

in the MARC-AMC fonnat and of an SAA convened

Thesaurus Study Group which is looking into

destinctions between physical fonnat and genre/fonn

of material terms.

EPHEMERA:

Fred Meyer Charitable Trust (Library & Infonnation

Resources for the Northwest, 1515 SW Fifth Ave., Suite

500, Portland, OR 97201) Moving fufonnation: Graphic

Images on CD-ROM, by David Miller, Nov. 1986, 35pp.

Interagency Committee on fufonnation Resource

Management (lAC/IRM), Minutes of Meeting of Nov. 12,

1986. Bill Hooton (NARA) identified three barriers tooptical disk projects becoming fully operational -

legal admissability of disk stored images, longevity

determination for medium, and data recording

standards.

National Archives & Records Administration,

Strategic Plan for Infonnation Systems & Technology,FIscal Years 1987-1991, Sept 1986, 168pp.

National Archives & Records Administration,

Archival Research & Evaluation Staff, General

fufonnation Summary, February 1, 1987, 5pp.

Nationallnfonnation Standards Organization (NISO

Z39), Subcommittee response to comments on Holdings

Statements for Non-Serial Items, draft standard Z39.57­

198x.

North Carolina Department of Cultural Resources,

General Description of FAIOS, March, 1987,22pp.

'Trends in Publishing: The Impact on Library

Storage", by Richard R. Rowe (President, The Faxon

Co.), presented to the Library Study Committee of the

Ohio Board of Regents. Planning a new building?

Rowe's reflections on library storage are worth

considering.

Society of American Archivists, Automated Records

& Techniques Task Force, Minutes of the Meeting of 26August, 1986, 15pp. & 1987 program proposals, sent

by Harold Naugler, dated February 5, 1987.

Society of American Archivists, Automation

Program Officer (Lisa Weber), Archival Automation

Infonnation Survey, Test Survey Fonn (December 22,

1986) for clearinghouse on automation in archives.

Fonns to be distributed in April.

voLl, #1

PLANS AND PROPOSALS:

ACTIONS VOCABULARYThe Wiltshire Museum Service Group in

England has developed a vocabulary ofacquisition, deposit and disposal actions whichis proposed as a starting point for atenninology control working group of theMuseum Documentation Association. Wnting ina recent issue of MDA Infonnation (v.l0,#I),Stuart Holm suggestS that the list, whichcontains tenns such as bequest. exchange, loan,offer for gift, purchase and transfer completewith scope notes and a syndetic structure,might serve the needs of a broader community,which he invites others to join by writing tothe MDA (Building 0, 347 Cherry Hinton Rd.,Cambridge, CB1-4DH, ENGLAND).

FRONT-END SYSTEMS:As Lenore Sarasan has rightly noted in a

stream of publications over the past decade,automation efforts in cultural repositoriesfrequently break: down around data entrybefore they payoff in retrieval. Good dataentry, data validation systems have rarely beenintegrated modules of the information retrievalsystems which attract archivists and museumcurators. In a several issues of MDAInformation; Richard· Light has.been reportingon MDA's internal development of amicrocomputer package consistent with MDAstandards which they hope to release in April1987. Last summer the Smithsonian Institutionissued an RFP for a similar system to serve as afront-end for all its bibliographic systems. DataEase software was selected and Richard Szaryand Avra Michelson expect to release a front­end for the S.I.Bibliographic InformationSystem (SIBIS) in April as well.

RESEARCH & EVALUATION:In a major contribution to the archiVal

profession and an insurance policy for its own .continued existence, the National Archivescreated the Archival Research & Evaluationstaff in January 1983. The range of questionswhich this relatively small group has beenable to address and the fundamental characterof the answers they are beginning to provide,should be better known. As reported elsewhere,they recommended an optical disk effort inFebruary 1984 which has been contracted. InOctober 1984 they completed a study of digital

raster scanning, optical character recognitionand speech pattern recognition technologies.In February of 1986 they introduced bar codelabels into a system to control the audio-visualinventory. Their study of the potential ofexpert systems in archives may be known tothe community through the publications ofDan DeSalvo in Telematics & Informatics, but .their on-going research into the use of opticalcharacter recognition devices for convertingtyped and handwritten indexes and fmding aidsare less publicized. Other projects includeresearch on preservation of paper, microfJ1mand magnetic media which could be of greatsignificance to the field. Office DirectorWilliam Holmes has agreed to keep me infonnedabout their work, and I look forward to havinghim and his staff report it here.

OVERCOMING OPTICAL LIMITS:The Council on Library Resources announcesthat Gary W. Collins is working with theLibrary of Congress to test optical disk agingand to detennine the conditions which inducedegradation of data stored on optical disks.

A committee co-chaired by Robert Wedgeworth(Dean of Columbia U. Library School) and· KurtD. Steele (Standard & Poor's) has proposedstandards for copyright protection of materialstransferred to optical disks. The standards,which affirm copyright across media, areexpected to have a significant impact onpublishers. Contact MaryBeth Peters, Registerof Copyright, Library of Congress, WashingtonDC 20540.

STATUS OF: THE AAT:by Toni Petersen

The Art and Architecture Thesaurus, a projectof the Getty Art History Infonnation Program,has been in operation since 1981. Its purpose isto provide a standardized vocabulary of art andarchitecture tenns· for use by bibliographic,archival, and visual databases and for theinventorying of object collections. It is hopedthat such standardization of input and retrievallanguage will result in. greater usersatisfaction and create a unified set of conceptdescriptors for indexers and researchers.

The AAT has built its vocabulary from suchsources as Library of Congress SubjectHeadings, RILA, and the Avery. Index toArchitectural Periodicals, augmented byliterature of the field drawn from dictionaries,

I

Spring 1987 Copyright by Archives & Museum Infonnatics 17

monographs. The terms are arrangedhierarchically (i.e., conceptually) andalphabetically. There is an attempt to provideaccess points via all known synonyms orvariants of a term. The MARC Format forAuthorities is the system used for transmittingthe AAT to users online. There will also be aprint version which can be periodicallyupdated.

AAT terminology is prepared by an editorialstaff according to ANSI and BSI standards.using the National Library of Medicine's MeSHthesaurus as a model. It is reviewed andapproved by advisory boards composed ofexperts in the areas covered

There are presently about 30,000 termscovering the following categories: DesignConcepts. Design Elements and Attributes.Styles and Periods. People and Organizations,Processes and Techniques. Disciplines andOccupations, Materials, ArchiteeteeturalComponents. Single Built Works. Complexes.Settlements and Landscape. Tools and Toolcomponents. Drawings, Document Types andVisual Genre. Further development isunderway in areas of the visual arts andmaterial culture.

A new computerized thesaurus maintenancesystem was installed recently consisting of ffiMATs networked to a Britton-Lee IntelligentDatabase Machine (IDM). utilizing a thesaurusconstruction program developed by the GettyArt History Information Program. The 10Menhances the previous system. providing fornumerous additional features such as theability to access any tenn across all of thehierarchies. A term authority record,containing full information on an AAT term -- its defInition, sources. variant parts of speech(e.g.• "etching" and its variant "Etched"). andfields for scope notes and related terms are alsoprovided. Work is underway to add both scopenotes and related terms to all existinghierarchies. The next project for the AAT isthe design of an improved display format forAAT terminology. and an online user systemwhich looks ahead to direct access to the AATby outside users.[Toni Petersen is Director of the AAT.located at62 Stratton Rd.. Williamstown, MA 01267]

STANDARDS:1987 has seen some extremely important

developments on the standards front ThisDepartment will regularly report on standardsissues which archivists should be aware of.

OPEN SYSTEMS STANDARDSThe National Bureau of Standards has issued

regulations which require all governmentagencies to adhere to the InternationalStandards Organization Open SystemsInconnection (ISO-OS!) model in purchasesfrom the summer of 1987 forward.

Last summer, the Department of Defensecommitted itself to making its TCP/IP family ofprotocols converse with OSI.

Major hardware vendors agreed late in 1986to revamp the Manufactwing AutomationProtocols (MAP) and the Office Protocols (TOP)which are major subsets of OSI in release 3 tobe available by 1988.

The Linked Systes Project Policy Committee(the project to link the Library of Congress,OCLC. RLG. and WLN) agreed on protocols whichare consistent with the ISO-oSI model fortransfer of records from networks to localsystems. They also announcved agreement onpriorities. including project implementation,developing the capability to exchangebibliographic records among participants andassuring that the local linkages used by all fourparticipants will be identical and LSP based.OCLC, the Library of Congress and RLG willimplement intersystem shared cataloging andinterlibrary loaning by the end of 1987. TheWestern Library Network did not announce ananticipated completion date.

The effect of these OSI developments is two­fold: First. archivists can assume that ISO-OSIwill be the direction of the future and shouldplan accordingly even though fullcompatibility based on the model still has manyhurdles to overcome. Secondly, as an effectivenational bibliographic system is welded. fromOCLC. RLG & WLN participation in one oranother network will increasingly bedetermined by the services which eachnetwork decides to offer. It will be easier forarchives to part company with their parentinstitutions primary bibliographic networkwhen such a course is dictated.

Fll..E INTERCHANGE STANDARDSThe American National Standards Institute

recently threw its support behind theInternational standard ISO-82I1. The standardspecifies medium independent. hardwareindependent and software independent fileand data formats which allow for the exchangeof mes between different computer systemswith a minimum of external specification. Thisis made possible because the envelope carries

18 Archival Infonnatics Newsletter Vol. 1, #1

along with the data, infonnation about how thedata is to be mapped into different structures toprevent infonnation loss. While the standardhas yet to affect software vendors (see reporton DBMS archiving issues by Thomas E. Brown& William A. Reader in this issue), many hopes(especially archival ones) depend on it Thespecification (32 p.) is available from ANSI(1430 Broadway, NY, NY 10018 for $9.00).

CATALOGING STANDARDSThe Joint Steering Committee for the

Revision of the Anglo-American CatalogingRules (AACR) 2nd edition agreed to the contentof a revision of those rules to be published inmid-1988. A new version of chapter 9, oncomputer files, will be published later thisyear.

The Library of Congress CatalogingDistribution Service has issued update 14 to theMarc Fonnats for Bibliographic Description.Also available is Betsy Betz Parkers' longawaited Thesaurus of Graphical Tenns.

The Society of American Archivists hassubmitted a grant proposal to NEH to enableSteve Henson to revise his Archives. PersonalPapers & Manuscripts: A Cataloging Manual.

VOCABULARY STANDARDSA meeting of the RLG working group on a

functions vocabulary ( attended by DavidBearman, Max Evans, Jamie Ingram, MarionMatters, Alden Monroe, Kathleen Roe, AlanTucker & Lisa Weber), reached an agreementabout how to proceed in the Seven-StatesProject exploration of the potential of"functions" tenninology to provide access tocommon activities across differentbureaucratic organizations.

Two levels of tenns were proposed for a test:a generic, "sphere of activity" tenn will beapplied to agency history records and a"function" tenn will be applied to series leveldescriptions of records. The participants will betesting these tenns in the spheres ofCorrections and Education over the next severalmonths.

A subcommittee of the RLG AMC Task Forceproposed standards and examples for recordingrecords schedules in RLIN. The intention ofthe standards advanced by Kathleen Roe andAlden Monroe is to pennit the Seven Statesproject to input some scheduling records to testthe potential of such infonnation for recordsmanagers and archivists.

TECHNICAL REPORT ON OPTICALMEDIA

Part 20f the Ouarterly Update on ArchivalInformatics: Newsletter & Technical Re.ports, isa technical report on optical media in archivesand museums. The following is a summary forthose readers who receive only the Newsletter.Non-subscription copies ofthe technical reportare available for $45.00 each.

Technologies which employ light to write andread information, while in their infancy, havetremendous poteQt:4ll for archives andmuseums. Laser-reflective optical videodischas·already demonstrated its value and is nowbeing used quite widely in collectionsdocumentation, exhibit and public infonnationapplications. CD-Audio has taken the consumermarket by storm, but its archival and museumapplicability is still uncertain. CD-ROM is nowsufficiently standard to support many datadistribution applications and we can expect tosee new products for the archives and museummarket emerging regularly over the nextseveral years. Optical digital disk is so new thatuntil very recently only national institutionslaunching prototype and demonstration effortscould claim to have demonstrated archival ormuseum applications, but in the past year anumber of other large ventures have beenlaunched, some of which have moved frompilots to production.

The attraction of each of these technologiesis different because their capabilities aredifferent. Laser reflective optical videodisc isextremely attractive for publishing images,especially segments of motion images, withsound and data in interactive fonnats. It is astable technology which boasts hundreds ofsuccessful implementations in archives andmuseums; the videodisc cost $2000 to master,copies are cheap and playback devices arestandard and available for under $1000. CD­ROM is very attractive for publishingtraditional databases, as evidenced by theinterest of the Census Bureau in using it for the1990 census, and can accommodate stilJ images;mastering costs are fairly high and playbackdevices are not yet standard peripherals, butboth can be expected to drop significantly soon.Optical digital disk is exciting because it is notlimited to being a publishing medium - it canbe directly updated by local devices. It alsoholds enonnous quantities of data and provides

(continued next page)

Spring 1987 Copyright by Archives & Museum Infonnatics 1987 19

OPTICAL MEDIA REPORT, continued from p. 19

rapid random access. Nevertheless, optical digital disk is a very non-standard technology and thecosts of systems are hard for most archival and museum applications to justify.

In this report the technologies themselves are examined in less detail (with references to morecomplete treatments) than are applications. Criteria for selecting projects for each technologyare discussed. as are cost factors associated with each and the staffing and support requirements ofprojects which use these technologies. Case studies of previously successful archives and museumcommunity projects are assessed. A series of checklists which can be used to determine whetheran application is suited to these technologies, and if so which technology and under whatassumptions, is provided for management guidance. Lists of current vendors and descriptions oftheir services are provided along with extensive bibliographies of background studies. casestudies and critical reviews.

To subscribe, please rill out or copy this form:

ARCHIVAL INFORMATICS NEWSLETTER & TECHNICAL REPORTS:Part 1 only: Archival Informatics Newsletter:

Enclosed is my check: $24 to U.S. address ($US 30. foreign)Please bill me: $40.

Parts 1 & 2: Newsletter & Technical Reports:Billed or pre-paid: $160.

Signature for all billed orders

>

Mail To:

Archives & Museum Informatics5600 Northumberland St.

Pittsburgh, PA 15217

Bill To: