ph: +1 416-691-2516
info @ archimuse.com
published: March 2004
Preservation Metadata for the Real World: The NC ECHO Preservation Metadata Model
Helen R. Tibbo, Kevin Cherry, and Claire Eager, University of North Carolina at Chapel Hill, USA
There is considerable pressure within the digital library, archive, and museum communities for institutions to create extensive, high-quality metadata for discovery, administration, structuring, and preservation of their digital assets. Large cultural heritage institutions and international collaborations are developing complex preservation metadata models, but it remains unclear how small and often under-staffed repositories will build extensive metadata databases for their digital objects and virtual collections. This paper reports on the efforts of the NC ECHO Preservation Metadata Working Group to create a model that includes information sufficient for the long-term preservation of digital entities, but which is simple enough for a wide range of institutions to use effectively and economically. The resulting NC ECHO Metadata for Preservation of Digital Images (MAPDI) data model and associated Access database are described herein.
Keywords: metadata, preservation
"Metadata is fashionable." (Baca, 2000)
For library, archival, and museum professionals, metadata is fashionable with presentations and articles on metadata appearing in most archival, library, and information science (LIS) conferences and professional journals and even in publications such as Scientific American (Rothenberg, 1995). While metadata has not yet been the subject of an MTV special or made the front page of USA Today or People magazine, the challenges of digital preservation that Hedstrom (1998) encapsulates so well have entered the public consciousness and the popular press (Bergerson, 2002; Kaplan, 1998; Manes, 1998; Schofield, 2003; Stille, 1999; Tangley, 1998).
There is considerable pressure within the digital library, archive, and museum communities for institutions to create extensive, high-quality metadata for discovery, administration, structuring, and preservation of their digital assets. Several models for these various types of metadata now exist, with emerging standards such as the Dublin Core (DC; www.dublincore.org ) for discovery metadata gaining widespread support and use.ï Lagging somewhat behind the development of descriptive metadata models, preservation metadata schemes, some of them quite complex, are now appearing internationally. To date, few but the largest of repositories such as the Library of Congress (www.loc.gov), Harvard University (www.harvard.edu), and the National Library of the Netherlands (www.kb.nl) have employed extensive metadata schemes to capture information essential to the preservation of their digital assets. It remains unclear at this time how smaller, less well-staffed repositories will build extensive metadata databases for their digital objects and virtual collections. This paper discusses the development of preservation metadata models in general and describes the efforts of the NC ECHO Preservation Metadata Working Group to create a model which contains sufficient information for the long-term preservation of digital entities, but which is simple enough for a wide range of institutions to use effectively and economically.
The Brief History of Preservation Metadata
Even though the term "metadata" is relatively recent in origin, its most-often cited definition, "data about data," is already considered by librarians and information scientists to be an "old saw," coming close to ranking alongside such phrases and concepts as "information age," "digital revolution," and "information explosion." Dempsey and Heery (1998) have noted that this "data about data" definition, nearing trite status, is "too terse to get us very far." (p. 145) While metadata can relate to any entity, physical or digital, and appear in any format from stone inscriptions to electronic database entries, Hudgins, Agnew, and Brown (1999) define the term in its most common usage to be "any data that aids in the identification, description and location of networked electronic resources." In addition to this "resource discovery function," Hudgins et al also note that metadata is data that helps with the control of an electronic resource, allowing a digital object's use to be tracked, its provenance to be maintained, and any of its rights and permissions to be managed. Dempsey and Heery (1998), along with Wendler (1999), set the stage for today's complex and extensive metadata schemes, dividing the subject into descriptive, structural and administrative metadata. It is within Hudgins' "control function" and Dempsey and Heery's "administrative component" that preservation metadata belongs. Preservation metadata is that set of information elements (resolution, compression, date of creation, etc.) about a digital resource that is collected and maintained to ensure its "viability, renderability, and understandability" over the long term. (OCLC/RLG Working Group on Preservation Metadata, 2002).
General Observations on the Need of Preservation Metadata
In May 1997, joining similar proposals beginning at least five years earlier (Bearman, 1992; Rothenberg, 1996) Michael Day made a call for using metadata as a tool to help manage the preservation of digital objects in an online or networked library setting. This paper and others like it were the preservation metadata equivalent of the "toe in the water." At this early point Day concluded his brief article with some very important questions, many of which remain unanswered: "Who will define what preservation metadata are needed? Who will decide what needs to be preserved? Who will archive the preserved information? Who will create the metadata? Who will pay for it?" While preservation metadata models are emerging from institutions and collaborations around the world, effective and widespread implementation of these schemes is yet to be a reality.
Specific Element Sets Supporting Major Methods of Preservation
Thibodeau (2002) provides an excellent overview of technical approaches to preservation of digital data, and Cloonan and Sanett (2000, 2002) review digital preservation activities at a number of institutions. Until recently, three preservation methods have received the most attention: 1) the computer museum model, 2) migration (Besser, 2001), and 3) emulation (Rothenberg, 1999; Granger, 2000). All of these methods and the others that Thibodeau discusses require the compilation of preservation metadata including information such as the version of the software used to create the digital objects, change/migration history, and various technical settings. (Day, 2000b) Or as David Bearman (1994) has argued, "content, structure and context information must be linked to software functionality that preserves their executable connections . . . representations of their relations must enable humans to reconstruct the relations that pertained in the original software environment." Bearman's "content, structure, and context information" and the "representations of their relations" mean preservation metadata.
In 1997, The Research Libraries Group (RLG) convened a seven-member working group to identify specific information elements to be captured to support digital preservation activities. The group began its final report, published in May 1998, by noting that up until then, metadata had been thought of primarily in terms of "resource discovery" Expanding the concept to include long-term maintenance issues, they proposed a preservation metadata element set, much of which could be gathered at the time of digital imaging. Even though the report was limited to converted digital images (no video, no audio, and no born-digital items), the group identified sixteen elements that they considered essential for the endurance of this one type of digital object. (RLG,1998) Many of these elements are contained within the more extensive preservation metadata models that have appeared since 2000.
By 1999, Michael Day could identify several other consortia and institutions involved in designating specific element sets to aid in their preservation work. Much of the work was then taking place in the archives/records management/electronic records realm. Among those institutions Day cited was the National Library of Australia (NLA), which developed "a logical data model to help identify particular entities and their associated metadata that needed to be supported within its PANDORA Project (Preserving and Accessing Networked Documentary Resources of Australia.)"(NLA, 1999). Day also discussed the University of Pittsburgh's "Functional Requirements for Evidence in Recordkeeping Project," which had developed metadata specification for evidence based upon a reference model for "Business Acceptable Communications" (BAC). The Pittsburgh project proposed a six-layer record structure, which would contain a "handle layer," or a resource discovery metadata component (Bearman & Sochats, 1996). Day also noted the work underway at the University of British Columbia, which had stepped back from the specific set of preservation elements to think about ways not only to preserve digital objects but also to ensure the authenticity of the records. The UBC project, relying upon one of the oldest of archival sciences, diplomatics, explored the authenticity of electronic records using theories and practices usually associated with those European government documents--more often than not--committed to vellum and parchment rather than CDs and servers. Out of this research, the UBC project developed a set of eight templates to form a metadata scheme. (Duranti, 1995; Duranti & MacNeil, 1996). This project led to the InterPARES Project, an international collaboration to study the long-term preservation of electronic records (www.interpares.org). Day also pointed out that Jeff Rothenberg, usually associated with emulation as preservation, joined Tora Bikson in proposing a preservation metadata framework for the Dutch National Archives and Ministry of the Interior in 1999 (Rothenberg & Bikson, 1999).
Additional element sets were on their way. The Networked European Deposit Library (NEDLIB) proposed its own set, as did the CURL Exemplars in Digital Archives Project (CEDARS). The NEDLIB Project addressed issues of technological obsolescence and attempted to be as generic as possible to allow for broad applicability. (Lupovici & Masanes, 2000) CEDARS, too, sought to be of use to a wide range of digital objects and did so "independent of level of granularity." (CEDARS, 2000; Day, 2000a) Across the Atlantic, researchers at Harvard's Digital Repository Service were exploring the use of the XML mark-up schema in its own list of preservation metadata elements (Harvard University). Harvard's use of XML enables its community to "deposit" digital objects into a controlled "datastore," complete with storage, access, and preservation functions. Community members leave digital objects in a "drop box" along with an associated instruction file in XML that notes the relevant metadata. The instruction file is then parsed, and the object is added to the datastore. There are several types of metadata needed for the system to function: global defaults (processing orders, etc.), base object metadata (regardless of type, i.e., image, sound, video, etc.), and type object metadata dependent upon type. The US National Archives is also developing its own electronic records preservation protocol, Electronic Records Archives (ERA) (NARA).
With continuing research and conferences focusing on digital preservation (e.g., Dale), it become apparent that standardization of preservation metadata element sets was essential. (Day, 1998) In March 2000, OCLC and RLG announced a collaborative project to develop a "comprehensive preservation metadata framework applicable to a broad range of digital preservation activity." According to Meg Bellinger, OCLC's Vice President for Digital and Preservation Resources, "The rationale for the work was based on the recognition that excellent work had been done in a number of test bed environments and what was needed was an authoritative synthesis of this before further work could be undertaken." (Kenny, 2001) In the resulting OCLC/RLG white paper (2001), the Working Group on Preservation Metadata assessed the current state of preservation metadata and found little consensus for best practice, but it did note several shared themes. Reviewing the work of NEDLIB, CEDARS, NLA (National Library of Australia), and the Harvard Digital Repository Service, the OCLC/RLG group found that a consensus between the varying models was feasible. They found noticeable and important points of convergence, a shared desire to create a technological independent element set (capable of supporting both emulation and migration), and agreement on the purpose of the various preservation endeavors (OCLC/RLG, 2001).
Each of the projects under review had its own approach to preservation and metadata and the OCLC/RLG group identified significant disparity among them. It found disagreement in several areas including: the level of specificity of preservation information required and level of granularity to which that information should be applied, interoperability of existing metadata standards and initiatives, and lack of agreed upon appropriate syntax or container to hold the preservation metadata element sets. (e.g., MARC or XML). In search of an organizing principle, the OCLC/RLG working group incorporated a high-level framework for the preservation of digital objects, the Open Archives Information System (OAIS).
The High-Level Framework: OAIS
In 1998, Allison Chilvers and John Feather proposed something that they called "supermetadata," data about metadata. For example, they could foresee some future distant date when someone, somewhere, would like to know the life history of the metadata associated with a particular digital object. In their investigation of this over-riding metadata, they found that the practitioners they interviewed shared questions that revealed "a desire to formulate some framework which will allow the information content of digital resources to be maintained. ... Solutions to these problems need to embrace a wide spectrum of stakeholders and their needs but at the same time be flexible enough to embrace future digital formats." Such a framework was to appear in OAIS.
OAIS grew out of the National Aeronautics and Space Administration (NASA)'s attempt to preserve the digital information generated by space exploration (CCSDS, 2002). A product of the Consultative Committee for Space Data Systems (CCSDS), an international body of space agencies, OAIS provides a model for digital archiving. Acronym filled (with DIPS and SIPS and AIPS), the model proposes an overall construct and vocabulary required for further, more detailed investigations into preservation of electronic information, specifying ingest, repository, and access architecture and activities. OAIS identifies four different types of preservation metadata that should be created for the Preservation Description Information (PDI) of the (Archival Information Package API): 1) reference information, 2) context information, 3) provenance information, and 4) fixity information. (CCSDS, 2002) The OCLC/RLG white paper noted that CEDARS, NLA, and NEDLIB preservation metadata schemas all map--more or less--to the four areas specified by the OAIS model. (The Harvard Digital Repository did not.) In June of 2002, the OCLC/RLG working group's final report made recommendations for implementation (specific metadata elements) for the two components of the OAIS framework directly related to preservation, Content Information and Preservation Description Information.
In fewer than ten years, thinking about preservation metadata has evolved from the general conviction that collecting and managing a certain set of information elements would aid in the long-term maintenance of digital information, to the testing of various sets of that information, to some agreement on a standard group of elements that fit under the umbrella of a generally accepted high-level model of digital archiving. One of the next steps for preservation metadata will be its incorporation back into the overall triumvirate of metadata: structural, administrative, and descriptive. One framework that facilitates this is the Metadata Encoding and Transmission Standard (METS), "a standard for encoding descriptive, administrative, and structural metadata regarding objects within a digital library, expressed using the XML schema language of the World Wide Web Consortium." (METS, 2001) METS is an initiative of the Digital Library Federation that builds from work in the California Digital Library (www.cdlib.org) and the Making of America II Project (http://sunsite.berkeley.edu/MOA2/). The Network Development and MARC Standards Office of the Library of Congress maintain the standard.
Another keenly important step will be the development of easy-to-use tools and templates that translate the research, studies, white papers, and conference Powerpoint presentations into activities that institutions, large and small, can readily accomplish while facilitating interoperability between individual institutions. As Hodge (2000) noted when making the case that preservation metadata is best created at a digital object's beginning, metadata is rarely captured at the creation because no tools have yet been forged to make the chore relatively easy. This is still the case, even though we have a much better idea what it is that we want those "to-be-forged" tools to capture.
NC ECHO Preservation Metadata Project
In April 2002, the State Library of North Carolina awarded the School of Information and Library Science at the University of North Carolina at Chapel Hill LSTA funds to develop metadata tools for its North Carolina ECHO (Exploring Cultural Heritage Online) program and statewide participants. NC ECHO is an evolving statewide collaborative that provides access to special collections and digitization projects. This project involved development of 1) a web-based template to allow for HTML mark-up of the Dublin Core metadata fields that NC ECHO could mount on its website to aid partner institutions with the incorporation of appropriate Dublin Core tags, and 2) a preservation metadata model with which NC ECHO participating repositories could capture metadata essential to the long-term preservation of their digital assets. The initial model presented here is customized to capturing preservation information about digital photographs but is adaptable to other types of primary source originals.
Purpose of the Model
Because NC ECHO works with a wide range of institutions and their staffs, many of which do not have advanced training in digital preservation or metadata creation and use, the most challenging aspect of this project was not to identify information necessary for digital preservation, but to keep the model and database simple enough so that staff and volunteers at all levels of repositories could easily use it. This demanded that we extract fields from some very complex models from leading projects around the world and develop explanatory documentation and examples to be distributed with the database from ECHO, either on CD-ROMs or in a downloadable version from the website.
The model and database do not represent all metadata that could be collected about digital images or more complex digital objects, nor are we sure that it collects every data point that will prove necessary for digital preservation in years to come. At this point no one knows precisely what information must be kept. Many of the large institutions around the world are hedging on the side of complexity and completeness. The Library of Congress, for example, is developing the quite complex, though flexible and extensible, METS framework, and Harvard University has an extensive preservation metadata schema for digital images. We consulted both of these models in our deliberations and drew heavily from them in our final product. In their totality, however, we considered them, and many others, to be too complex for staff at most repositories, even the larger ones within North Carolina, to use or afford. Clearly, a good deal of metadata must be kept if digital objects are to survive repeated migrations and other preservation transformations, but each piece of metadata kept extracts a significant cost from the repository. It was our goal, therefore, to develop a model and database that staff could easily understand and use and that would not consume an unrealistic amount of time and resources to compile. Presumably, tools to automatically capture creation metadata (scanner settings, image dimensions, etc.) will bring down the cost of metadata creation in the future, but for the small repository these may remain out of economic and technical reach for some time to come.
In April 2002 the NC ECHO Preservation Metadata Working Group was formed to develop the intellectual model that would underlie the database. The working group consisted of :
Development of the Metadata Model
The development and implementation of standards is central to the longevity of any digital files and objects. We started our work with the premise that any metadata model, guidelines, or recommendations we would make for NC ECHO would fall within the framework of evolving national and international standards. As noted above, much of the work in preservation metadata is being conducted within the structure of the Open Archival Information System (OAIS) initiative and developing a model that would be OAIS compliant was one of our initial decisions.
We identified several projects and organizations that had developed or were using metadata models that were compliant with the OAIS framework and extracted relevant data elements from them. Projects we reviewed included :
The Working Group extracted data elements common to most of these models that are necessary for long-term preservation of digital images. The result is the Metadata for Preservation of Digital Images (MAPDI) data element list and associated Access database described below. We tested the model and database with a small number of NC ECHO participants for clarity and usability, but more extensive evaluation based on actual implementations should be conducted.
System Design Considerations
MAPDI gives users the ability to add, edit, and delete records describing digital images. In most cases, one record represents one digital image. It also has several builtiin reports that give information about a specific image, or overview information about an entire digital collection.
MAPDI was designed for the express purpose of serving repositories whose resources may be limited, either in actual funds or in technical training and support. Microsoft Access was selected for its ease of use, immense user support community, and availability to state employees who make up a large portion of the NC ECHO repositories. MAPDI was built using the bare essentials of MS Access. The intent of this Spartan development approach was to facilitate further customization of the system by each repository if they so desire. Additions can be made with minimal threat of interrupting existing processes, and the learning curve for any developer expanding the system will be extremely low. It does result in a system that is not as feature-rich as it could be, but it is fully functional on delivery and allows for the easy capture of preservation metadata. Should a repository obtain grant funding or technical assistance, as we hope many will, many small but significant changes could be made to enhance the system with a moderate amount of expertise and minimal cost.
The element set consists of 30 elements and seven element-specific qualifiers. The elements cover creation of the digital image, image identification, image properties, and rights metadata. Clearly, each element was included because it was deemed important to long-term maintenance of the image, but it is understood that not all repositories will have the time, training, or resources to capture all the information covered in the element set. For this reason, only six elements were designated as "Required," and a "Strongly Suggested" designation was added to indicate those elements that should be populated if a repository is able. Only the Revision History element is repeatable.
There is a reason why metadata is fashionable. It is this "data about data" that makes access to digital information and its long-term preservation possible. Great strides have been made during the last several years in the development of preservation metadata element sets, but this is only a necessary first step. A great deal is left to be done in educating web designers, electronic publishers, and cultural institution online developers in the use of these element sets. Even more important work lies in providing the tools that will allow the day-to-day practitioners to readily and consistently capture preservation metadata.
The NC ECHO Preservation Metadata Working Group strove to balance the need for extensive preservation metadata and the realities of the repositories that will be ECHO participants. At this time it is unclear if the model we developed is too complicated or too time consuming to implement for many repositories. Only actual use of the database, not merely user testing to see if the interface is clear and appropriate, will tell us if it is realistic to expect a wide range of repositories to gather and maintain extensive preservation metadata for images or any of their digital objects.
What is clear is that managing the digital objects and providing means of preservation and access through technical approaches and metadata are significant challenges to fully implemented and functional projects and programs. The small repository must be able to build and maintain metadata databases for its digital assets, or these must be centralized in the ECHO structure. Most of the over 800 cultural heritage repositories in the state of North Carolina lack any technical support. Data management software must be easy to use and require little customization. For a statewide, distributed model such as NC ECHO to succeed in its goal of bringing the rich cultural heritage of the state to the digital arena, the collaborative must support participants with easy-to-use information technology solutions and expertise.
Baca, M. (2000). Introduction to metadata: Pathways to digital information. Los Angeles: Getty Research Institute.
Bearman, D. (1992). Documenting documentation. Archivaria 34, 33-49.
Bearman, D. (1994). Electronic Evidence: Strategies for managing records in contemporary organizations. Pittsburgh, Archives and Museum Informatics.
Bearman, D. and K. Sochats (1996). Metadata requirements for evidence. Pittsburgh, University of Pittsburgh, School of Information Science. Consulted November 15, 2002. http://www.archimuse.com/papers/nhprc/BACartic.html.
Bergerson, B. (2000). Dark ages II: When the digital data die. Upper Saddle River, NJ: Prentice Hall PTR.
Besser, H. (2001). Digital preservation of moving image material? The Moving Image. Consulted January 15, 2003. http://www.gseis.ucla.edu/~howard/Papers/amia-longevity.html
CCSDS (2002) Reference model for an open archival information system (OAIS). Recommendation for space data system standards. CCSDS 650.0-B-1, January 2002. Consulted January 15, 2003. http://www.ccsds.org/documents/650x0b1.pdf.
CEDARS ( 2000). Metadata for digital preservation: The CEDARS Project outline specification. Consulted November 15, 2002. http://www.leeds.ac.uk/cedars/MD-STR~5.pdf
Chilvers, A. and J. Feather (1998). The management of digital data: A metadata approach. Electronic Library 16, 365-372.
Cloonan, M. V. and S. Sanett (2002). Preservation strategies for electronic records: Where are we now: Obliquity and squint? American Archivist 65, 70-106.
Cloonan, M.V. and S. Sanettt (2000). Comparing preservation strategies and practices for electronic records. New Review of Academic Librarianship 6, 205-216.
Colorado Digitization Project. Consulted January 15, 2003. http://www.cdpheritage.org/resource/index.html.
Dale, R. (2000). Digital preservation conference: Report from York, UK. RLG DigiNews 4/6. Consulted January 15, 2003. http://www.rlg.org/preserv/diginews/diginews4-6.html#feature2
Day, M. (1997). Extending metadata for digital preservation. Ariadne May 1997. Consulted January 15, 2003. http://www.ariadne.ac.uk/issue9/metadata
Day, M. (1998). Issues and approaches to preservation metadata, Joint RLG and NPO Preservation Conference: Guidelines for digital imaging. Scarman House, University of Warwick, Coventry, September 1998. Consulted on January 15, 2003. http://www.ukoln.ac.uk/metadata/presentations/rlg-npo/warwick.html
Day, M. (1999). Metadata for digital preservation: An update. Ariadne 22. Consulted January 15, 2003. http://www.ariadne.ac.uk/issue22/metadata/
Day, M. (2000a). Metadata for preservation (CEDARS Project Documentation AIWO1). Consulted on January 15, 2003. http://www.ukoln.ac.uk/metadata/cedars/AIW01.html.
Day, M. (2000b). Resource Discovery, Interoperability, and Digital Preservation: Some Aspects of Current Metadata Research and Development. Vine 117, 35-48. Consulted January 15, 2003. http://www.ukoln.ac.uk/metadata/publications/vine-117/.
Dempsey, L. and R. Heery (1998). Metadata: A current view of practice and issues. Journal of Documentation 54, 145-172.
Duranti, L. (1995). Reliability and authenticity: The concepts and their implications. Archivaria 39, 5-10.
Duranti, L. and H. MacNeil (1996). The protection of the integrity of electronic records: An overview of the UB-MAS research project. Archivaria 42, 46-67.
Granger, S. (2000). Emulation as a digital preservation strategy. D-Lib Magazine 6/10. Consulted January 15, 2003. http://www.dlib.org/dlib/october00/granger/10granger.html.
Harvard University. Harvard's Digital Repository Service's website. Consulted January 15, 2003at.http://hul.harvard.edu/ois/systems/drs.
Hedstrom, M. (1998). Digital preservation: A time bomb for digital libraries. Consulted January 15, 2003. http://www.uky.edu/~kiernan/DL/hedstrom.html.
Hodge, G. (2000). Best practices for digital archiving: an information life cycle approach. D-Lib Magazine 6/1. Consulted January 15, 2003. http://www.dlib.org/dlib/january00/01hodge.html.
Hudgins, J., G. Agnew, & E. Brown (1999). Getting mileage out of metadata: Applications for the library. Chicago: American Library Association. Consulted January 15, 2003. http://www.lita.org/litapubs/lg5.html.
Kaplan, K. (1998). The culture's immortal but not the disk. Los Angeles Times, February 16, 1998, D, 1:4.
Kenney, A. R. (2001). Editor's interview: Collaboration of RLG and OCLC with digital archiving issues. RLG DigiNews 5/6. Consulted January 15, 2003. http://www.rlg.org/preserv/diginews/diginews5-6.html
Manes, S. (1998). Time and technology threaten digital archives. New York Times, April 7, 1998, F4.
METS. (2001). Metadata Encoding and Transmission Standard (METS) Official Website. METS: An overview and tutorial. Consulted January 15, 2003. http://www.loc.gov/standards/mets/METSOverview.html
NARA. Electronic Records Archives. http://www.archives.gov/electronic_records_archives/index.html
National Library of Australia (1999). Preservation metadata for digital collections. Consulted January 15, 2003. http://www.nla.gov.au/preserve/pmeta.html
Lupovici, C. and J. Masanes. (2000). Metadata for long-term preservation of electronic publications. NEDLIB Report Series 2. Consulted January 15, 2003. http://www.kb.nl/coop/nedlib/results/NEDLIBmetadata.pdf
OCLC/RLG Working Group on Preservation Metadata (2001). Preservation metadata for digital objects: A review of the state of the art: A white paper. Consulted January 15, 2003. http://www.oclc.org/research/pmwg/presmeta_wp.pdf
OCLC/RLG Working Group on Preservation Metadata (2002). Preservation metadata and the OAIS reference model: A metadata framework to support the preservation of digital objects. Consulted January 15, 2003. http://www.oclc.org/research/pmwg/pm_framework.pdf.
RLG. Working Group on Preservation Issues of Metadata (1998). Final report. Consulted January 15, 2003. http://www.rlg.org/preserv/presmeta.html.
Rothenberg, J. (1995). Ensuring the longevity of digital documents. Scientific American 272, 42-47. While Rothenberg does not use the term "metadata" specifically, he talks about the need to store system specifications for the emulation of old software, that is, metadata about the software.
Rothenberg, J. (1996). Metadata to support data quality and longevity. Proceedings of the first IEEE Metadata Conference, April 16-18,, 1996. Consulted January 15, 2003. http://computer.org/conferences/meta96/rothenberg_paper/ieee.data-quality.html
Rothenberg, J. (1999). Avoiding technological quicksand: finding a viable technical foundation for digital preservation. Washington, DC: Council on Library and Information Resources. Consulted January 15, 2003. http://www.clir.org/pubs/reports/rothenberg/contents.html
Rothenberg, J. and T. Bikson (1999). Carrying authentic, understandable, and usable digital records through time: Report to the Dutch National Archives. Consulted January 15, 2003. http://www.digitaleduurzaamheid.nl/bibliotheek/docs/final-report_4.pdf.
Schofield, J. (2003). Digital dark age looms. Guardian January 9, 2003, 1-3. Consulted January 15, 2003. http://www.guardian.co.uk/online/story/0,3605,871091,00.html
Stille, A. (1999). Overload. The New Yorker 75 March 8, 1999, 38.
Tangley, L. (1998). Whoops there goes another CD-ROM. U.S. News and World Report, 124/6 February 16, 1998, 67.
Thibodeau, K. (2002). Overview of technological approaches to digital preservation and challenges in coming years. In The state of digital preservation: An international perspective. Conference proceedings. Institute for Information Science, April 24-25, 2002, Washington, DC. Washington, DC: Council on Library and Information Resources. Consulted January 15, 2003. http://www.clir.org/pubs/reports/pub107/thibodeau.html.
United States National Archives and Records Administration (NARA), Electronic Records Archives. Consulted January 15, 2003. http://www.archives.gov/electronic_records_archives/index.html
Wendler, R. (1999) LDI update: Metadata in the library. Library Notes 1286, 4-5.