7
Archives and Museum Informatics 11: 261–267, 1997. 261 c 1997 Kluwer Academic Publishers. Printed in the Netherlands. Records Management Research Sponsored by the US Army and the Department of Defense* WILLIAM E. UNDERWOOD Artificial Intelligence Atlanta, Inc., Decatur, GA 30030 USA Abstract. This paper is a synopsis of results achieved in addressing research questions and issues in the management of electronic records. The formulation of records management theory within the context of set theory is briefly described. Steps toward establishing functional requirements for electronic recordkeeping are summarized. The use of software prototyping to develop software techniques for classifying, profiling, and records disposition is described. New research issues and approaches to their resolution are proposed. The methods used in this research suggest the importance of interdisciplinary collaboration in addressing issues in the management of electronic records. Key words: archival theory, electronic records, functional requirements, recordkeeping, records management theory 1. Introduction This paper discusses research in records management sponsored by the Army Research Laboratory (ARL) at the Georgia Institute of Technology and the Depart- ment of Defense (DoD) Records Management Task Force. An objective of this research has been to demonstrate the use of commercial-off-the-shelf (COTS) recordkeeping software in supporting classification, filing, retrieval and disposition of electronic records according to MARKS, the integrated records classification, filing and disposition schedule of the US Army. This has involved reengineering the Army’s integrated file plan and records schedule and manual filing system for paper records to operate with the COTS electronic recordkeeping system. The objective has been generalized to address electronic recordkeeping needs in all DoD agencies. The question of functional requirements has been pursued in col- laboration with the School of Library, Archival and Information Studies (SLAIS) of the University of British Columbia (UBC). This research is supported by the Army Research Laboratory (ARL), Computer Systems Tech- nology Division (CSTD), Atlanta, GA, and the DoD Records Management Office under contract DAKF1-93-C-0043 (OCT 93-July 1997) to Artificial Intelligence Atlanta, Inc., Decatur, GA. The findings in this article are not to be construed as an official Department of Army or Department of Defense position unless so designated by other authorized documents.

Records Management Research Sponsored by the US Army and the Department of Defense

Embed Size (px)

Citation preview

Page 1: Records Management Research Sponsored by the US Army and the Department of Defense

Archives and Museum Informatics 11: 261–267, 1997. 261c 1997 Kluwer Academic Publishers. Printed in the Netherlands.

Records Management Research Sponsored by theUS Army and the Department of Defense*

WILLIAM E. UNDERWOODArtificial Intelligence Atlanta, Inc., Decatur, GA 30030 USA

Abstract. This paper is a synopsis of results achieved in addressing research questions and issuesin the management of electronic records. The formulation of records management theory withinthe context of set theory is briefly described. Steps toward establishing functional requirementsfor electronic recordkeeping are summarized. The use of software prototyping to develop softwaretechniques for classifying, profiling, and records disposition is described. New research issues andapproaches to their resolution are proposed. The methods used in this research suggest the importanceof interdisciplinary collaboration in addressing issues in the management of electronic records.

Key words: archival theory, electronic records, functional requirements, recordkeeping, recordsmanagement theory

1. Introduction

This paper discusses research in records management sponsored by the ArmyResearch Laboratory (ARL) at the Georgia Institute of Technology and the Depart-ment of Defense (DoD) Records Management Task Force. An objective of thisresearch has been to demonstrate the use of commercial-off-the-shelf (COTS)recordkeeping software in supporting classification, filing, retrieval and dispositionof electronic records according to MARKS, the integrated records classification,filing and disposition schedule of the US Army. This has involved reengineeringthe Army’s integrated file plan and records schedule and manual filing systemfor paper records to operate with the COTS electronic recordkeeping system. Theobjective has been generalized to address electronic recordkeeping needs in allDoD agencies. The question of functional requirements has been pursued in col-laboration with the School of Library, Archival and Information Studies (SLAIS)of the University of British Columbia (UBC).

� This research is supported by the Army Research Laboratory (ARL), Computer Systems Tech-nology Division (CSTD), Atlanta, GA, and the DoD Records Management Office under contractDAKF1-93-C-0043 (OCT 93-July 1997) to Artificial Intelligence Atlanta, Inc., Decatur, GA. Thefindings in this article are not to be construed as an official Department of Army or Department ofDefense position unless so designated by other authorized documents.

VICTORY PIPS: 145147 ARMUarmu27.tex; 20/08/1997; 23:28; v.7; p.1

Page 2: Records Management Research Sponsored by the US Army and the Department of Defense

262 WILLIAM E. UNDERWOOD

2. Progress, Unanswered Questions and Unresolved Issues

In this section, progress achieved by this project on research questions and issuesare summarized. New questions and research issues are raised and approaches totheir answers and resolution are suggested.

Research question: Can logical theories of archival and records managementconcepts be formulated within the context of set theory?

This question is motivated in part by the need for generic models of recordsclassification, records filing and retrieval, and retention scheduling. In other words,not just examples of these, or schemes of a particular style, but a generic model bywhich one can prove theorems that are true for each of these functions.

Another motivation for this research question is that in defining a businessprocess model for recordkeeping, when one considers the classification, filing,retrieval or disposition activity, the operations (controls) and data structures requiredfor that operation are defined by the theory. It is easier to say what the functionalrequirements of a particular activity (function) are if there is already a preciselogical theory of that function within the context of set theory. It is also easier toprogram the function.

Logical theories of business activities, records classification, filing and retrieval,and records disposition have been formulated and some interesting theoremsproven.1 One of the theorems implies that if an agency and offices within anagency have a well-structured business activity model in which the creator of arecord is able to associate a record with the business activity, then a functionalrecord classification scheme is not needed for record classification. Such a classi-fication scheme might be of use for retrieval of non-current records by businessactivity, for retention, or for archival description, but its not needed for recordclassification.

Another implication of these theories is that it is not necessary to includethe attributes of reliable records in the core metadata of records to ensure theirreliability. Those attributes are manifested in, and a part of, the records themselves.

These initial results suggests that this is an important research question thatshould continue to be pursued. It raises research issues such as:

– Can a logical theory of the authenticity of records be formulated within thecontext of set theory?

– Can a logical theory of principles such as Respect du Fonds and the conceptof aggregation in archival description be formulated within the context of settheory?

– Can logical theories of business and documentary procedures be formulatedwithin the context of set theory?

1 Underwood, W., Formal Models of Business Activities and Records Classification, Filing andRetention AIAI TR 97-04, ARL, CSTD (Atlanta, GA: Georgia Institute of Technology, 1997).

armu27.tex; 20/08/1997; 23:28; v.7; p.2

Page 3: Records Management Research Sponsored by the US Army and the Department of Defense

RECORDS MANAGEMENT RESEARCH 263

Positive resolution of such research issues could result in an organized body ofknowledge of records management and archives that could be more readily appliedto practical issues in such areas as formulation of records management policy,development of recordkeeping systems, and records migration.

Research question: What are the functional requirements for electronicrecordkeeping? What are the information and data requirements for electronicrecordkeeping?

If electronic records are to be managed by an electronic recordkeeping system,software developers need to know what functions they must implement and whatdata must be maintained. If an organization decides to purchase a commerciallyavailable electronic recordkeeping system, it must be sure that the system performsthe functions and maintains the data required by the organization. A functionalrequirement is a statement of a function which a system must implement. A datarequirement is a statement of the logical data structures that must be created, used,and maintained by the system.

The DoD Records Management Task Force has reengineered the process ofmanaging electronic records in the DoD.2 The IDEF0 and IDEF1X modelingmethodologies were used to develop a TO BE business activity model and datamodel for managing electronic records.3 A statement of baseline functional require-ments for a records management application (RMA) was developed from thesemodels. The functional requirements were formulated as a standard for softwareacquisition.4

A research project conducted by the School of Library, Archival and Informa-tion Studies, UBC has addressed the development of functional requirements fora recordkeeping system that preserves the integrity of the electronic records. Col-laboration with the DoD Records Management Task Force involved adopting theIDEF0 and IDEF1X methodologies and software tools that support these method-ologies. From the agency perspective, a business activity model and data modelwere developed for Managing the Archival Fonds.5

2 Thibodeau, K. and Prescott, D., “Reengineering Records Management: The DoD Records Man-agement Task Force,” Archivi and Computer 6(1) (1996).

3 US Department of Commerce, Federal Information Processing Standards Publication 183,Integration Definition for Function Modeling (IDEF0), December 21, 1993. US Department ofCommerce, Federal Information Processing Standards Publication 184, Integration Definition forInformation Modeling (IDEF1X), December 21 1993.

4 Department of Defense, Standard for a Records Management Application, DoD 5015.XX-F(Draft) (www.dtic.dla.mil/c3i/recmgt).

5 Duranti, L., MacNeil, H., and Underwood, W., “Protecting Electronic Evidence: A SecondProgress Report on a Research Study and its Methodology”, Archivi & Computer 6(1) (1996): 37–69.U.S. Department of Defense, Records Management Program Management Office and Universityof British Columbia, SLAIS, Applying IDEF Methodology to Describe Archival Science, Report 5,October 30, 1996.

armu27.tex; 20/08/1997; 23:28; v.7; p.3

Page 4: Records Management Research Sponsored by the US Army and the Department of Defense

264 WILLIAM E. UNDERWOOD

The DoD models designed by the Records Management Task Force were devel-oped by records management practitioners in the DoD. The UBC/DoD modelswere developed by people conversant with the principles of archival science anddiplomatics.6 Functional requirements for electronic recordkeeping have not yetbeen developed from the UBC/DoD models, so the research question is not yetanswered.

Archival science and diplomatics are among the controls on the business activ-ities of the IDEF0 activity model. A conscious attempt has been made to justifyprocedures and data elements in the model in terms of archival principles anddiplomatics. To a certain extent, the business activity and data models are mod-els of the theories of diplomatics and archival science. That is, they are a partialinterpretation of the theories.

One of the issues that arose during this research is “To ensure the reliability ofrecords, must records of transactions and communications external to an agencybe registered and the register preserved, even when records are destroyed?” Inthe United States, when records can be destroyed and are destroyed, one doesn’tmaintain a register summarizing the basic elements of the record. In some countries,e.g., Italy, the register is retained even when the original record is destroyed, toensure accountability of the government agency. In the Italian juridical system, anagency’s records cannot be used against it in litigation. Not so in the United States.Should a capability to register external records be a functional requirement of anelectronic recordkeeping system? If so, what should the retention period be? Theseare important issues in the United States.

Research issue: Can a methodology be defined for developing functionalrequirements for electronic recordkeeping from an IDEF0 business activity modeland an IDEF1X data model for a business activity such as “Manage ArchivalFonds”?

IDEF0 and IDEF1X are not integrated methodologies. It is possible to derivea model for a business activity that is not coherent with a data model for thesame activity. Furthermore, together they do not provide a complete methodologyfor defining functional requirements. Nor is there any other IDEF methodology, ofwhich there are about a score, for defining functional requirements. These issues arebusiness process reengineering issues, not archival and records management issues.But their resolution is a prerequisite to the question of functional requirements ofelectronic recordkeeping, if adopting the IDEF methodologies.

6 Duranti, L., “Diplomatics: New Uses for an Old Science,” Archivaria (1988–1992), Part I, 28:7–27; Part II, 29: 4–17; Part III, 30: 4–20; Part IV, 31: 10–25; Part V, 32: 6–24; Part VI, 33: 6–24.

armu27.tex; 20/08/1997; 23:28; v.7; p.4

Page 5: Records Management Research Sponsored by the US Army and the Department of Defense

RECORDS MANAGEMENT RESEARCH 265

A solution to this research issue has been developed.7 Briefly, to supportfunctional requirements definition, IDEF0 must require something comparable torule-based definition of controls on the lowest-level business activities in the model.A method is suggested for defining IDEF1X data models based on the informationreferenced in the rules defining the lowest-level activities of the IDEF0 activi-ty models. The definition of IDEF1X data models requires an iterative return tothe IDEF0 activity model to ensure the rules defining the lowest level activitiesreference all and only the data elements in the data model. Then a method is sug-gested for defining functional requirements from the rules. The plan is to developfunctional requirements from the UBC/DoD models using this methodology.

3. Software Prototyping as a Method for Addressing Research Issues

In this section, records management research issues and results are discussedthat involve software prototyping. Rapid (software) prototyping also can be usedas a method for determining functional requirements for an information system.This approach to requirements definition involves demonstrating the function tobe performed, and obtaining the concurrence of an organization that this is thefunction that is needed. Then a statement of the function is abstracted from thebehavior of the prototype that does not commit one to a specific procedure.

Research issue: How can creators of records be assisted in classifying a recordaccording to a functional classification scheme?

This research question was originally motivated by the fact that officers in Armyoffices must assign a file code to records that they receive or create. The file codemust be from an office file list which is constructed from approximately 3000MARKS filing categories. Records are then filed by this file code. Retrieval andproper disposition depend on assignment of this file code. People have a great dealof difficulty assigning these codes, due in part to the imprecision of the descriptionsof filing categories.

This issue was investigated by developing a well-defined records classificationscheme, and an automated filing assistant that had the knowledge to interpretthe record and to interact with the officer to classify the record. Syntactic andsemantic models of functional records classification scheme were developed.8 Itwas demonstrated that an intelligent filing assistant could support the recordscreator in classifying paper, e-mail, word processing and scanned documents in

7 Underwood, W., Duranti, L., Prescott, D. and Kindl, M., “Extensions of IDEF Methodologybased on US DOD CIM Experience in Reengineering the Records Management Process,” WorldMulticonference on Systemics, Cybernetics and Informatics, Sci’97: International Symposium inBusiness Process Reengineering, Caracas, Venezuela (1997).

8 Underwood, W., The Language of Records Classification, AIAI TR 97-02, ARL, CSTD (Atlanta,GA: Georgia Institute of Technology, 1997).

armu27.tex; 20/08/1997; 23:28; v.7; p.5

Page 6: Records Management Research Sponsored by the US Army and the Department of Defense

266 WILLIAM E. UNDERWOOD

close to 100% of the cases.9 This is accomplished easily and very quickly. Thisperformance is better than human performance on the same task. Note that this isnot automated classification, but assistance to the officer who makes the decisionas to the proper classification.

Research issue: To what degree is automated support of electronic recordsdisposition possible?

This question was motivated by the fact that in US Army offices and other FederalAgencies, someone has to determine from generic disposition instructions forrecord series, the specific disposition dates or events for record series. Someonehas to type these specific instructions on labels of file folders or enter them into anelectronic recordkeeping system. They have to do this at least annually. Someonehas to check these record series at least annually and apply the specific dispositioninstructions, not only to paper records but electronic records and records on othermedia. This is a tedious, time consuming job that often is delayed due to a shortageof office administrative personnel.

The disposition instructions of MARKS and NARA’s General Records Schedulewere analyzed to determine the types of generic disposition instructions. A gram-mar was developed that defines all and only the language of disposition instructions.The semantics of generic disposition instructions was defined. A syntax checkerand translator were developed. 10 It was demonstrated that when files and folders(dossiers, cases) are created in an office, the specific disposition instructions for therecord series containing the file can be automatically generated. The person autho-rized to carry out the disposition instructions is notified of the disposition actionon the occurrence of the disposition date or event. They are supported in cuttingoff, transferring, or destroying electronic and other records.11 These grammars andtools suggest the possibility of a standard form for retention schedules and dispo-sition instructions along with guidelines and automated tools for constructing andapplying them.

Research issue: To what degree is automated profiling and filing of electronicrecords possible?

This question is motivated in part by the need to identify attributes of records foruse in subsequent retrieval. Personnel do not want to have to identify attributes ofa record needed in the profile.

Analysis of MARKS filing categories resulted in the discovery that there aredifferent indexing and retrieval criteria for different document types. Furthermore,

9 Underwood, W. and Laib, S., Evaluation of a Reengineered MARKS in Managing ElectronicRecords, ARL, CSTD (Atlanta, GA: Georgia Institute of Technology, 1997).

10 Underwood, W. and Laib, S., The Language of Records Disposition, AIAI TR 95-01, ARL,CSTD (Atlanta, GA: Georgia Institute of Technology, 1995).

11 Underwood and Laib, Reengineered MARKS in Management of Electronic Records.

armu27.tex; 20/08/1997; 23:28; v.7; p.6

Page 7: Records Management Research Sponsored by the US Army and the Department of Defense

RECORDS MANAGEMENT RESEARCH 267

for electronic records other than standardized forms, it was desirable to defineelectronic document templates for different document types that identify fieldscontaining values for the attributes needed in the profile.

To a certain extent this has already been accomplished for electronic mail.Attributes that are needed for managing electronic mail as electronic records areprovided as attributes in the header of the electronic mail. Electronic mail is anexample of an encapsulated object.

Software prototypes that automatically profile electronic mail, automatically fileit according to MARKS, and that support retrieval based on these attributes havebeen demonstrated. Retrieval of these records for different combinations of attribut-es, and even the information content of the records, also has been demonstrated.12

It can be concluded that electronic templates for different document types are agood solution to the record profiling issue.

4. Conclusion

It has been shown that logical theories are a method of organizing knowledge ofrecords and archival management. This is a promising approach to establishingtheoretical foundations of archives and records management. Unanimity amongarchivists as to the basic principles of records and archival management wouldfacilitate the development of such formal theories. The development of such formaltheories also should contribute to greater unanimity among archivists as to thefoundations of their discipline.

The question of functional requirements for electronic recordkeepng is unan-swered. However, progress has been make in collaborative research of UBC, DoDand ARL. The methodology formulated for defining functional requirements fromIDEF0 and IDEF1X models should contribute to a solution to this research ques-tion. Continued sponsorship and collaboration is needed to carry this work tocompletion.

New software techniques have been demonstrated for records classification,profiling and records retention. This suggests that the functional requirements forelectronic recordkeeping can contain functions performed by the system that werepreviously performed by office personnel.

The research results summarized in this paper were based on process and datamodeling methodology, the axiomatic method and software prototyping. These arenot the customary methods of archival and records management studies. If theresearch questions and issues and the results discussed in this paper are appropriateto the discipline, interdisciplinary research is probably required on those questionsand issues.

12 Underwood and Laib, Reengineered MARKS in Management of Electronic Records.

armu27.tex; 20/08/1997; 23:28; v.7; p.7