D-Lib Magazine
November 2003

Volume 9 Number 11

ISSN 1082-9873

Using MPEG-21 DIDL to Represent Complex Digital Objects in the Los Alamos National Laboratory Digital Library

Jeroen Bekaert
Los Alamos National Laboratory, Research Library and Ghent University, Faculty of Engineering
[email protected]

Patrick Hochstenbach
Los Alamos National Laboratory, Research Library
[email protected]

Herbert Van de Sompel
Los Alamos National Laboratory, Research Library
[email protected]

Abstract

Various XML-based approaches aimed at representing complex digital objects have emerged over the last several years. Approaches that are of specific relevance to the Digital Library community include the Metadata Encoding and Transmission Standard (METS), the IMS Content Packaging XML Binding, the Sharable Content Object Reference Model (SCORM), and the XML packaging approach developed by CCSDS Panel 2. The MPEG-21 Digital Item Declaration Language (DIDL) is another XML-packaging specification that, so far, has received little attention in the Digital Library community. This article gives a brief insight into the MPEG-21 standardization effort, and indicates its potential relevance to the Digital Library community. It also highlights major characteristics of DIDL, and details research conducted at the Research Library of the Los Alamos National Laboratory (LANL) into the applicability of DIDL for the representation of complex objects in the LANL repository. The positive outcome of this research has led to a decision to make DIDL-conformant documents the unit of storage in the LANL repository, and suggests that DIDL could also be a valuable building block for other Digital Library projects.

1.   Introduction

Digital Libraries have reached a point where acceptable architectures must accommodate objects that aggregate datastreams of a wide variety of media-types, and must allow for the association of secondary data – including metadata supporting discovery, digital preservation and rights management – with those datastreams. Also, the notion of the – static or dynamic – association of dissemination capabilities with such objects must be entertained by contemporary Digital Library architectures.

The seminal Kahn/Wilensky framework [10] refers to a digital object as the basic entity for storing, managing and disseminating information in a Digital Library architecture. The framework defines a digital object as more than just a set of bits: it should be regarded a data structure with a unique persistent identifier that, apart from the datastream(s), holds secondary information about the contained datastream(s). It comes as no surprise that, since the publication of the Kahn/Wilensky framework, various approaches have emerged from different communities aimed at representing digital objects. Approaches that are of specific relevance to the Digital Library community include the Metadata Encoding and Transmission Standard (METS) [30], the IMS Content Packaging XML Binding [8], the Sharable Content Object Reference Model (SCORM) [1] and the XML packaging approach developed by CCSDS Panel 2 [5]. All these efforts specify an XML-based data structure for digital objects.

Through collaboration with the Multimedia Lab of Ghent University [note 6], the authors became aware of another XML-based packaging format that emerged from the ongoing MPEG-21 standardization effort [3]. The MPEG-21 work has received little attention in the Digital Library community, possibly because specifications are not freely available online, as a result of an ISO strategy. MPEG Standards must be purchased from ISO [note 1], while reference software is made publicly available. Digging deeper into MPEG-21 documentation, the authors became increasingly intrigued, and decided to actively explore the applicability of MPEG-21 concepts in the context of ongoing Digital Library repository work at the Research Library of the Los Alamos National Laboratory (LANL). The main motivations for this decision can be summarized as follows:

• Potential impact of MPEG-21 – MPEG [note 2] is an ISO/IEC Committee, and provides a mechanism to feed research results into an ISO standard. So far, several MPEG Standards have had a significant impact on the multimedia landscape. For example, MPEG has produced the MPEG-1 and MPEG-2 Standards, on which formats such as Video CD, MP3, Digital Television, and DVD are based. It is likely that the MPEG-21 Standard will have a similar impact.

• MPEG-21's modular architecture – The vision for MPEG-21 is ‘to define a normative open framework for multimedia delivery and consumption for use by all the players in the delivery and consumption chain’. The envisioned framework covers the entire media content delivery chain, encompassing content creation, adaptation, and dissemination. The unfinished Standard consists of 12 high-level, modular Parts, including:
• MPEG-21 Part 2 – Digital Item Declaration Language (henceforth referred to as DIDL), detailing the representation of complex digital objects [15]
• MPEG-21 Part 3 – Digital Item Identification Language (henceforth referred to as DII), detailing the identification of complex digital objects and their contained entities [17]
• MPEG-21 Part 4 – Intellectual Property Management and Protection (henceforth referred to as IPMP), detailing a framework to enforce rights expressions pertaining to complex digital objects [21].
• MPEG-21 Part 5 – Rights Expression Language (henceforth referred to as REL), detailing a language to express rights pertaining to complex digital objects and their contained entities [22]
• MPEG-21 Part 7 – Digital Item Adaptation (henceforth referred to as DIA), detailing the adaptation and transcoding of datastreams based on contextual information such as agent capabilities, network characteristics and user preferences [18]
• MPEG-21 Part 10 – Digital Item Processing (henceforth referred to as DIP), detailing the association of processing methods with complex digital objects and their contained entities [19, 20]

• Ability to accommodate any media type and genre – Although MPEG-21 originates in a community that focuses on motion picture, audio, and video, the MPEG-21 framework can accommodate any kind of complex digital objects including electronic texts, electronic journals, scientific datasets, etc.

• Applicability to Digital Libraries – There is a clear overlap between the problem domain addressed by the MPEG-21 effort, and ongoing efforts regarding the representation, storage, and dissemination of complex digital objects in the Digital Library community. For example, MPEG-21 DIDL and DII directly relate to the aforementioned XML-based packaging approaches. Also, MPEG-21 DIP and DIA reveal a remarkable parallel with sophisticated architectures that emerged from the Digital Library community [23], specifically FEDORA [7, 26, 29] and SODA [12, 24]. And, recently, the usage of MPEG-21 REL has been proposed in the ISO/IEC SC36 E-Learning standardization effort [9].

This article describes the results of research into the applicability of the MPEG-21 DIDL and DII for packaging complex digital objects to be submitted to, stored in, and disseminated from the repository of the LANL Research Library. A future article will report on the repository architecture used to store and disseminate the complex digital objects, which builds on DIDL [note 3], the OAI-PMH [11], the NISO OpenURL Framework [25] and concepts from the MPEG-21 DIP.

2.   DIDL: MPEG-21 Digital Item Declaration Language

In the MPEG-21 Framework, complex digital objects are declared using the Digital Item Declaration Language (DIDL). DIDL introduces a set of abstract concepts that, together, form a well-defined data model for complex digital objects [note 4]. Based on those abstract concepts, DIDL defines a W3C XML Schema [note 5] that provides broad flexibility and extensibility for the actual representation of compliant complex digital objects. In the remainder of this article, a complex digital object represented according to the DIDL Schema will be referred to as a Digital Item Declaration, or DID. Core characteristics of the DIDL data model, the DIDL XML Schema, and DIDs are described in the remainder of this section.

2.1   Data Model

This section provides an informal, and simplified explanation of the DIDL data model. The DIDL data model recognizes the following entities, which are visually represented in Figure 1:

• A Container is a grouping of Containers and/or Items. In the XML representation, a Container is accommodated by the didl:Container element ({1} in Figure 1 and Appendix A)
• An Item is a grouping of Items and/or Components. In the XML representation, an Item is accommodated by the didl:Item element ({2} in Figure 1 and Appendix A)
• A Component is a grouping of Resources. Multiple Resources in the same Component are considered equivalent and consequently an agent may use any one of them. In the XML representation, a Component is accommodated by the didl:Component element ({3} in Figure 1 and Appendix A)
• A Resource is an individual datastream. In the XML Schema, a Resource is accommodated by the didl:Resource element ({4} in Figure 1 and Appendix A)
• Secondary information pertaining to a Container, an Item, or a Component can be conveyed by means of a Descriptor. In the XML representation, a Descriptor is accommodated by the didl:Descriptor element ({5} in Figure 1 and Appendix A). By definition, a didl:Descriptor is associated with its parent element in the XML representation. For example, a didl:Descriptor provided as a child element of a didl:Component is associated with that didl:Component.

The DIDL specification provides abstract definitions for each of the aforementioned entities and their interrelations. Those definitions can hardly be used as a cookbook for representing a collection of related datastreams as a DID; they actually allow for various approaches to do so. For example, when representing an audio album using DIDL, one could create a DID which has an Item per audio track, or a DID with a single Item containing multiple Components, etc. In many cases, the actual choice for one representation or another will be inspired by the requirements of the target application. Figure 1 shows the structure of a DID that is conformant with the DIDL specification. The figure clearly illustrates the elaborate nesting capabilities of DIDL.

Figure 1: Example of a DID structure conformant with the DIDL specification

2.2   Providing datastreams and secondary data

Table 1 shows the techniques available in DIDL to deliver datastreams and secondary information. As can be seen, DIDL allows these to be contained in a DID – By Value – or to be pointed at from within a DID – By Reference. Table 1 also shows how the nature of the datastreams and secondary information relates to the way in which they are provided in a DID. The triangles in Table 1 indicate functionality that most likely will become available as a result of amendments proposed by the authors and their colleagues from Ghent University. The MPEG Committee has recently approved these amendments [2] as a Working Draft for the DIDL Extension specification [16].

Table 1: By Value and By Reference provision of datastreams and secondary information in a DID

2.3   More about Descriptors

Descriptors provide an extensible mechanism to convey secondary information about entities of the DIDL data model. For example, in order to associate – say – an identifier with an Item, a didl:Descriptor containing the identifier can be created as a child element of the didl:Item element.

As will be shown in the remainder of this section, the MPEG-21 framework itself defines ways to use Descriptors as a means to convey identification information, rights information, processing information, etc. But, to facilitate the provision of community-specific or application-specific information, Descriptors may also be defined by third parties. In order to do so, typically, an XML Schema with an associated XML Namespace is created to contain elements and attributes required to address specific needs. This approach is illustrated in Section 3.3 by detailing the Descriptors defined at LANL.

2.3.1   DII: Using a Descriptor for Identification

MPEG-21 Digital Item Identification (DII) specifies the usage of Descriptors for the identification of DIDs and their contained entities. It introduces a DII XML Namespace with elements that can be used to associate identifiers and/or types with Containers, Items, Components and Descriptors. For example, Table 2 shows the use of the dii:Identifier element to associate the URI “urn:isbn:0-395-36341-1” with a didl:Item. Through the introduction of a special Part dedicated to the identification of entities, MPEG-21 recognizes the importance of identifiers in network-based applications. Some packaging approaches, including METS, address identification in less fundamental and less extensible manners.

2.3.2   Using Descriptors to convey processing information

MPEG-21 Digital Item Processing (DIP) specifies an architecture pertaining to the dissemination of DIDs. This Part – yet to be standardized – introduces a special type of Item named Processing Item, which may be used to specify the methods by which a DID and its contained entities can be processed. Items are identified as Processing Items by including a Descriptor containing the dii:Type element from the DII XML Namespace, and by assigning it the reserved value “urn:mpeg:mpeg21:2002:01-DIP-NS:PI”. It is worthwhile noting that the MPEG-21 Processing Item concept is closely related to Fedora’s “behavior” concept [7, 26, 29]. A Processing Item is physically contained in the same DID as the entity of the data model with which it is associated. For example, a Processing Item can be associated with a Container, an Item, or a Component. Because DIP remains to be standardized, a more detailed description of Processing Items is out of the scope of this article. It suffices to say that, typically, a Processing Item either contains or points at code that can be used to process an entity of a DID.

An entity of the data model can be accorded a dip:ObjectType element from the DIP XML Namespace. A dip:ObjectType associated in this manner with an entity can be used as a key to actual bind the entity to a Processing Item. As a matter of fact, this approach conveys that the processing method specified by the Processing Item uses as argument the entity of the DID for which the content of the dip:ObjectType element is the same as the content of the Processing Item’s dip:Argument. It is helpful to think of the dip:ObjectType as a ‘link’ between an entity and a Processing Item, not as an indication of the ‘genre’ of an entity.

Table 3 shows an Item associated with a Processing Item by means of the value ‘urn:my:Argument'. That value is used as the content of both the dip:ObjectType element of the Item, and the dip:Argument element of the Processing Item. It is also shown that the Processing Item is assigned the identifier 'urn:bar:a333766936' through the use of the dii:Identifier element from the DII XML Namespace. The DIP specification allows entities to have more than one dip:ObjectType. Also, a Processing Item can bind to more than one entity by using multiple dip:Argument elements, each of which connects via the dip:ObjectType to the entities.

2.3.3   Using Descriptors to convey rights expressions

MPEG-21 Rights Expression Language (REL) specifies the usage of Descriptors to associate rights expressions with DIDs and their contained entities. This is achieved through the introduction of a language inspired by XrML [6] with elements and attributes in a REL XML Namespace. Table 4 shows the use of the r:license element to associate simple copyright information with a didl:Item. MPEG-21 Intellectual Property Management and Protection (IPMP) will provide tools to enforce rights expressions declared by means of REL.

3   Usage of MPEG-21 DIDL at the LANL Research Library

When researching the usage of the MPEG-21 Digital Item Declaration Language to represent complex digital objects in the repository at the LANL Research Library, two major questions emerged:

• How to map the – related – datastreams to be contained in a complex object of the LANL repository to the DIDL data model
• How to use Descriptors to meet the design goals of the repository and its associated applications

Both questions will be addressed in the remainder of this section, by providing an insight into the major design choices that were made when implementing DIDL at the LANL Research Library. The described design choices result in a DID profile, which has formally been expressed as a Schematron schema [note 3]. DIDs conformant to the design choices must be valid according to the DIDL XML Schema as well as according to this Schematron schema.

In the remainder of this article, XML excerpts of a LANL DID are provided to illustrate the main design choices that were made. A full representation of the DID from which excerpts are taken is provided in Appendix A. This DID is the DIDL-based representation of a complex object consisting of:

• A LANL technical report, which is a single PDF file with identifier ‘urn:bar:99-7537’.
• Descriptive metadata about the LANL technical report expressed by means of the MARC format. Actually, two versions of the MARC data are provided: the original MARC record, and a MARCXML representation derived from it. The identifier of the MARC record is ‘urn:bar:56-8730’.

3.1   LANL DIDs grow in breadth, not in depth

DIDs at LANL make no use of the extensive nesting capabilities provided by the DIDL data model or the associated XML Schema. All DIDs use a simple 3-level hierarchy Container / Item / Component. As a result, the DIDs cannot grow in depth, but can do so in breadth through the addition of Items to a Container, Components to an Item, and Resources to a Component. Figure 2 illustrates this approach. As will be shown, hierarchical relationships that could be expressed by nesting entities are instead represented by means of a special purpose Descriptor.

Figure 2: Example of a LANL DID structure following the LANL DIDL profile

3.2   All LANL data is created equal

The LANL repository harbors tens of millions of records from abstracting and indexing databases. These metadata records are "stand-alone" in that they do not come with their full-content counterparts. As a result, when embedding such a record in a DID, it can hardly be provided as secondary data because there is no primary datastream to which to attach it. Therefore, it is provided as a datastream in its own right. This approach is generalized to also embed descriptive metadata as an autonomous datastream in a DID when the DID contains both the descriptive metadata and the content it describes. While this might be contrary to the mainstream approach in this respect, a case in its favor can be made. First, descriptive metadata – in many cases expensive to create – should also be the subject of digital preservation, and as a result be treated as a potentially endangered datastream in its own right. Second, as technologies evolve and datastreams of a variety media-types become directly searchable, the special status of descriptive metadata as the sole point of entry to those datastreams might eventually weaken, turning descriptive metadata and datastreams into peers. As will be shown, the relationship between the metadata and the content it describes is represented by means of a special-purpose Descriptor.

Table 5 shows a DID that represents the LANL technical report. The Container accommodates 2 Items. The first Item contains the MARC metadata about the LANL technical report. This Item contains two Components: the second has a Resource that is the - base64 encoded - original MARC record, the first has a Resource that is a MARCXML version of the original MARC record. The second Item contains the technical report provided by means of two Resources in a single Component, indicating the equivalence of both Resources. In the first Resource, the PDF is provided By Value, and hence is base64-encoded. For illustrative purposes, the PDF is also provided By Reference in a second Resource through the inclusion of a reference to its network-location.

3.3   LANL's usage of Descriptors

3.3.1   Identifiers

In Digital Library applications, identifiers are of utmost importance. Identifiers were at the core of the seminal Kahn/Wilensky Framework [10], and they rightfully received special attention in the OAIS model [28]. Therefore, identifiers became a core element in the design of LANL DIDs, and their implementation is based on MPEG-21 DII:

• Identifiers are mandatory for the identification of the DIDs themselves. These ‘DID-identifiers’ (see Figure 2) are provided as the content of the dii:Identifier element in a didl:Descriptor which has the sole didl:Container element of the DID as its parent. These identifiers are dynamically assigned when DIDs are created, and they are UUID URNs [13].
• To map datastreams to the DIDL data model, a strong relationship has been defined between Items and identifiers:
  • When a single datastream has an identifier, it must be treated as an Item.
  • When multiple datastreams share an identifier, their combination must be treated as an Item.
  • All Items must have an identifier.
  These ‘Content-identifiers’ (see Figure 2) are provided as the content of the dii:Identifier element, which, in this case, is contained in a didl:Descriptor child element of the didl:Item element. Multiple 'Content-identifiers' for a single Item can exist. As is the case with 'DID-identifiers', 'Content-identifiers' can be dynamically assigned when DIDs are created. Also, at LANL, ‘Content-identifiers’ are in many cases inferred from the data to be contained in the Item. For example, when a record from the PubMed database is included in a DID, it receives a 'Content-identifier' such as 'info:pmid/14577066', where 14577066 is the unique PubMed number for the record, and 'info' is a proposed URI scheme [31].

This identifier-centric approach to XML-packaging of datastreams is somehow in contrast with the hierarchy-centric perspective of approaches such as METS and IMS, in which structural metadata are at the core of the document models (cf. the mandatory structural map in METS). As will be shown, in LANL DIDs, structural and relational metadata is represented by means of a special-purpose Descriptor.

Table 5 shows the use of both the ‘DID-identifiers’ and ‘Content-identifiers’. A dii:Identifier element associates the ‘DID-identifier’ “urn:uuid:10ba6842-ec45-3b19-8kub-hy8ff58c58a8b” with the didl:Container element. And, a dii:Identifier is used to attach the ‘Content-identifiers’ ‘urn:bar:56-8730’ and ‘urn:bar:99-6537’ with the MARC record and the technical report, respectively.

3.3.2   Processing Items and their 'Placeholders'

In LANL, DIDs will be used as a mean to store and disseminate complex digital objects. The authors felt a general level of discomfort with embedding Processing Items - which express the methods that can be used to process contained entities – in stored DIDL objects. This discomfort is related to the anticipated need to frequently update the content of Processing Items as new processing methods emerge, or existing ones are updated. This anticipated need to regularly “touch” stored DIDs is a poor fit with the rather static nature of the contained content at LANL, and with the OAIS-inspired strategy to create new DIDs – instead of updating existing ones – when some form of editing has been performed.

As a result, it was decided not to embed Processing Items in DIDs, but instead to embed ‘Placeholders’ for Processing Items. When the dissemination of a stored DID is requested, the contained ‘Placeholders’ will dynamically be exchanged for actual processing-related information. The ‘Placeholder’ concept is implemented by means of a didl:Descriptor that contains a diph:PlaceHolder element from a self-defined XML Namespace. It can be attached at the didl:Container, didl:Item and didl:Component level. When a stored DID is disseminated, the following is achieved by looking up the content of each embedded diph:PlaceHolder element in a special-purpose registry:

• Based on correspondences expressed in the registry, a diph:PlaceHolder element is replaced by one or more dip:ObjectType elements
• Based on registry-information, Processing Items are added to the DID, and connected to the inserted dip:ObjectType elements in the manner described in Section 2.3.2.

The result of this approach is a dynamic way of binding stored DIDs to specific dissemination methods. These methods, however, are not hard-coded into the DIDs, and hence they can easily evolve over time. The ‘Placeholders’ can be regarded as a flexible bridge between an Archival Information Package and a Dissemination Information Package.

Table 6 shows the use of the diph:PlaceHolder element to include ‘Placeholders’ at the level of the Container, Item and datastreams. Table 7 shows the result of replacing the diph:PlaceHolder element of one of the Components with a corresponding dip:ObjectType and Processing Item. The correspondence between a diph:PlaceHolder element, the dip:ObjectType element, and the Processing Items is retrieved from the special-purpose registry. It should be noted here that one diph:PlaceHolder element may be replaced by multiple dip:ObjectType elements.

3.3.3   Relationships

A special-purpose Descriptor is introduced to express relationships between entities contained in DIDs. This Descriptor is based on a self-defined Digital Item Relations (DIR) XML Namespace, which contains RDF Statements expressing relations such as “isDerivationOf”, “isPartOf”, “isTranslationOf”, “isDescriptiveMetadataOf”, etc.

• RDF [34] is used as the language to express the entities involved in a relationship and the nature thereof.
• The RDF statements are based on a vocabulary of relationships. This vocabulary remains to be defined, and terms will probably be created when required. It may be possible to import terms from existing XML Namespaces, such as dcterms [4] and PRISM [27].
• For the identification of resources involved in relationships, the following approach is taken:
  • A DID is identified by means of its DID-identifier
  • An Item and Component are identified by means of an XML fragment identifier - XPointers [33, 35] - expressed relative to the DID. The actual nature of these XPointer expressions is the subject of ongoing research. One approach is to base XPointers on ID-typed elements. IDs are a robust identification method, as they remain valid even after repositioning an element in a document. However, this approach requires the introduction of IDs for every entity involved in a relationship. Another approach is to use XPointers based on the DID structure itself. This approach comes with little initial overhead, but processing XPointers of this type may be more cumbersome than processing ID-based XPointers.
• The following conventions were introduced to include relationship information in a DID:
  • Relationships between DIDs, and between a DID and resources external to the actual DID are expressed by attaching a DIR Descriptor to the corresponding didl:Container element.
  • Relationships between an Item and other Items contained in a DID, and between an Item and the Components within that Item, and relationships between an Item and resources external to the actual DID are expressed by attaching a DIR Descriptor to the corresponding didl:Item element.
  • Relationships between Components within an Item, and between Components and resources external to the actual DID are expressed by attaching a DIR Descriptor to the corresponding didl:Component element.
  These conventions are inspired by the distinct treatment of Context Information and Representation Information in the OAIS model. The OAIS Model defines Context Information as information that documents the relationships of a Data Object (i.e., the identified digital object) to its environment and how it relates to other Data Objects. Representation Information describes the internal structure (including hierarchies) of the Data Object. As such, it seems that relational information pertaining to the Items resorts under the OAIS Context Information, while relational information pertaining to Resources resorts under OAIS Representation Information. Pragmatism also played a role in deciding upon these conventions, because - from an application-perspective - it is easier to manipulate Items that contain their own relationships as this makes the Items self-contained.

Table 8 shows a Descriptor, describing the metadata/content relationships between the MARC record and the technical report described by the MARC record. It also shows the relationship between the Items and a larger collection. Because these are interrelations between Items and relations between Items and an external resource, the Descriptor is attached to the didl:Item element of the MARC record.

3.3.4   Creation date

A special-purpose Descriptor is introduced to express the datetime of creation of the DID entities contained in DIDs. This Descriptor contains a didt:Created element from a self-defined Digital Item DateTime XML Namespace. The date and time values are formatted according to the W3C profile of ISO 8601 [32]

Table 9 shows a didl:Container with a Descriptor that conveys the datetime of creation of the Container structure, namely ‘2003-09-05T21:51:01Z’, and a didl:Item with a Descriptor that accommodates the datetime of creation of the Item structure, namely ‘2003-09-05T18:30:07Z’. The Descriptor is also used to convey the datetime of creation of an actual datastream.

4   Conclusion

This article has described the application of the MPEG-21 Digital Item Declaration Language to represent complex digital objects from the collection at the LANL Research Library. Although the article has not explicitly touched on the matter, the authors hope to have shown that, generally, usage of MPEG-21 DIDL as a complex object document model for Digital Library applications is a feasible, and actually attractive option:

• From a strategic perspective, DIDL is appealing because it is part of the MPEG-21 suite of Standards which is likely to receive strong industry backing. Also, DIDL is part of a broader architecture that is relevant to many communities, including the Digital Library community.
• From a functional perspective, DIDL is attractive because of the flexibility offered by its well-specified data model, and because of the extensibility provided by the Descriptor approach. The MPEG-21 Standard itself makes use of those Descriptors to provide a fundamental solution for the identification of entities, to associate processing methods with entities, to express rights related to entities, and to allow for the enforcement of those rights. As has been shown by means of the LANL DIDs, Descriptors can be used to meet specific design requirements. At LANL, Descriptors have been used to enforce an identifier-centric document model, to implement a dynamic association between entities and processing methods, and to introduce a novel way to express relationships. More generally, Descriptors can be used as a tool to address community-specific interoperability requirements. For example, one of the authors has done initial research regarding the use of DIDL for the representation of complex digital objects for the purpose of digital preservation, using Descriptors as a technique to map OAIS metadata categories to the DIDL data model.

DIDs created at LANL are fully compliant with the DIDL specification. Compliance with the LANL DID design is enforced by first validating DIDs against the DIDL XML Schema, and next against a Schematron schema that formalizes the design characteristics of LANL DIDs. The DIDs will be the unit of storage in the LANL repository. Soon ingestion will start, and millions of DIDs will be created. A forthcoming article will describe the characteristics of the repository architecture used to store and disseminate DIDs. In that architecture, the OAI-PMH, the OpenURL and concepts from the MPEG-21 Digital Item Processing specification play a fundamental role.

Acknowledgements

The research reported in this article was supported by the Belgian Science Foundation (Section Flanders) and the LANL Research Library.

The authors want to thank:

• Their colleagues of the Prototyping Team of the Research Library of the LANL Laboratory for their invaluable input in this research: Luda Balakireva, Henry Jerez, Xiaoming Liu and Thorsten Schwander. Rick Luce at the Research Library for making this research possible and for continued encouragement. Miriam Blake and Beth Goldsmith from the Development Team of the LANL Research Library for valuable feedback regarding the LANL DID.
• Thomas DeMartini at ContentGuard Inc. for his helpful input on the use of MPEG-21 Rights Expressions in the LANL DID.
• Sandy Payette at Cornell University, Michael Nelson at Old Dominion University and Tony Hammond at Elsevier Science for reviewing a draft of this article.

Jeroen Bekaert also wishes to thank:

• His supervisors Prof. Rik Van de Walle (Multimedia Lab, Ghent University) and Prof. Mil De Kooning (Dept. of Architecture and Urbanism, Ghent University) for their ongoing support and encouragement.
• His colleagues of Multimedia Lab for sharing their MPEG-21 knowledge.

References

1. Advance Distributed Learning, “The Sharable Content Object Reference Model (SCORM) - Version 1.3 - WD,” March 2003

2. Bekaert, J., Hochstenbach, P., De Neve, W., Van de Sompel, H. and Van de Walle, R. “Suggestions concerning MPEG-21 Digital Item Declaration,” ISO/IEC JTC1/SC29/WG11 M9735, Trondheim, July 2003.

3. Burnett, I., Van de Walle, R., Hill, K., Bormans, J., and Pereira, F., “MPEG-21: Goals and Achievements,” IEEE Multimedia, Oct-Dec 2003, pp. 60-70.

4. Dublin Core Metadata Initiative, Dublin Core, http://dublincore.org/

5. Giaretta, D. CCSDS Panel 2, “Steps towards CCSDS standards for information packaging using XML,” December 2001.

6. Contentguard, XrML - eXtensible rights Markup Language, http://www.xrml.org/

7. FEDORA, “Mellon Fedora Technical Specification, Version 1.1,” http://www.fedora.info/, December 2002.

8. IMS Global Learning Consortium, “IMS Content Packaging XML Binding - version 1.1.2 - Final specification,” 2001.

9. Information Technology for Learning, Education, and Training, “Liaison Statement on MPEG-21 Rights Expression Language (MPEG-21 REL),” ISO/IEC JTC 1/SC 36/WG 4 N0052, Seoul, September 2003.

10. Kahn, R. and Wilensky R. “A Framework for Distributed Digital Object Services. Corporation for National Research Initiatives,” http://www.cnri.reston.va.us/k-w.html, 1995.

11. Lagoze, C., Van de Sompel, H., Nelson, M. and Warner, S. “The Open Archives Initiative Protocol for Metadata Harvesting - Version 2.0”, http://www.openarchives.org/OAI/2.0/openarchivesprotocol.htm, 2003.

12. Maly, K., Nelson, M. and Zubair, M. “Smart Objects, Dumb Archives: A User-Centric, Layered Digital Library Framework,” <doi:10.1045/march99-maly>, D-Lib Magazine, March 1999.

13. Mealling, M., Leach, P. and Salz, R. “A UUID URN Namespace”, http://www.ietf.org/internet-drafts/draft-mealling-uuid-urn-01.txt

14. MPEG-21, Information Technology, Multimedia Framework, “Part 1: Vision, Technologies and Strategy,” ISO/IEC TR 21000-1:2001, December 2001.

15. MPEG-21, Information Technology, Multimedia Framework, “Part 2: Digital Item Declaration,” ISO/IEC 21000-2:2003, March 2003.

16. MPEG-21, Information Technology, Multimedia Framework, “Part 2: Digital Item Declaration Extensions - Amendment 1 - WD v.1,” ISO/IEC JTC1/SC29/WG11 N5837, Trondheim, July 2003.

17. MPEG-21, Information Technology, Multimedia Framework , “Part 3: Digital Item Identification,” ISO/IEC 21000-3:2003, March 2003.

18. MPEG-21, Information Technology, Multimedia Framework, “Part 7: Digital Item Adaptation,” ISO/IEC JTC 1/SC 29/WG 11 N5845, Trondheim, July 2003.

19. MPEG-21, Information Technology, Multimedia Framework, “Part 10: MPEG-21 Digital Item Processing,” ISO/IEC JTC1/SC29/WG11 N5855, Trondheim, July 2003.

20. MPEG-21, Information Technology, Multimedia Framework, Part 10: MPEG-21 Digital Item Processing, “Technologies under Consideration (TuC) v.1,” ISO/IEC JTC1/SC29/WG11 N5622, Pattaya, March 2003.

21. MPEG-21, Information Technology, Multimedia Framework, “Part 4: MPEG-21 Intellectual Property Management and Protection,” ISO/IEC JTC1/SC29/WG11 N5874, Trondheim, July 2003.

22. MPEG-21, Information Technology, Multimedia Framework, “Part 5: MPEG-21 Rights Expression Language,” ISO/IEC JTC1/SC29/WG11, N5838, Trondheim, July 2003.

23. Nelson, M., Argue, B., Efron, M., Denn, S. and Pattuelli, M.C. “A Survey of Complex Object Technologies for Digital Libraries,” NASA/TM-2001-211426, http://techreports.larc.nasa.gov/ltrs/PDF/2001/tm/NASA-2001-tm211426.pdf, December 2001.

24. Nelson, M. “Buckets, smart objects for digital libraries, Ph.D. Dissertation,” Old Dominion University, August 2000.

25. NISO AX Committee, “The OpenURL Framework for Context-Sensitive Services,” Proposed Standard, http://library.caltech.edu/openurl/Standard.htm, 2003.

26. Payette, S. and Lagoze, C. “Flexible and Extensible Digital Object and Repository Architecture,” Second European Conference on Research and Advanced Technology for Digital Libraries, http://www.cs.cornell.edu/payette/papers/ECDL98/FEDORA.html, Heraklion, September 1998.

27. PRISM, Publishing Requirements for Industry Standard Metadata, http://www.prismstandard.org/

28. Space data and information transfer systems, “Open Archival Information System, Reference Model,” ISO 14721:2003, February 2003.

29. Staples, T., Wayland, W. and Payette S. “The Fedora Project: An Open-Source Digital Object Repository System,” <doi:10.1045/april2003-staples>, D-Lib Magazine, April 2003.

30. The Library of Congress, “Metadata Encoding and Transmission Standard,” http://www.loc.gov/standards/mets/, February 2002.

31. Van de Sompel, H., Hammond, T., Neylon, E., and Weibel S.L. “The "info" URI Scheme for Information Assets with Identifiers in Public Namespaces”, http://www.ietf.org/internet-drafts/draft-vandesompel-info-uri-00.txt

32. W3C, “Date and Time Formats,” Note, http://www.w3.org/TR/NOTE-datetime, September 1997.

33. W3C, “Proposal for XML Fragment Identifier Syntax 0.9,” Working Group Note, http://www.w3.org/TR/xml-fragid/, September 2003.

34. W3C, “RDF, Resource Description Framework,” Recommendation, http://www.w3.org/RDF/, August 2003.

35. W3C, “XML Pointer Language,” Recommendation, http://www.w3.org/TR/xptr/, August 2002.

Notes

1. ISO: http://www.iso.ch/

2. MPEG Website: http://mpeg.telecomitalialab.com/

3. LANL DID: http://lib-www.lanl.gov/proto/MPEG-21/

4. MPEG-21 DIDL Working Draft: http://xml.coverpages.org/MPEG21-WG-11-N3971-200103.pdf

5. MPEG-21 DIDL Working Draft Schema: http://download.webct.com/public/ims/2.0/MPEG21.xsd

6. Multimedia Lab: http://multimedialab.elis.ugent.be/

Appendix A: LANL DID representing a technical report and metadata describing it

Top | Contents
Search | Author Index | Title Index | Back Issues
Previous Article | Next Article
Home | E-mail the Editor

D-Lib Magazine Access Terms and Conditions

DOI: 10.1045/november2003-bekaert