Dr V.I. Britchkovski

Belarussian State University of Informatics and Radiolelectronics

An enormous amount of valuable digital information has been and is being is being amassed across the globe. These information resources have great potential for combination, reuse and exploitation when solving ecological probl ems. One obstacle to use and reuse derives from the fact that these resources are distributed across a heterogeneous environment which makes discovery and combination difficult, another is that quality, reliability and accuracy vary across different inf ormation sources.

Some intelligent mechanisms are needed to facilitate the identification, location, proper use and reuse of digital information resources. Such information resources are distributed across a heterogeneous environment, with logically re lated information stored in different locations, on different platforms, in different formats and with different software and access mechanisms. Quality, reliability and accuracy varies across different information sources.

The existence of mechanisms to intelligently classify, sort, filter and retrieve information in a focused and pertinent way is prerequisite to the effective exploitation and reuse of these information resources. The development and ad option of standardised metadata schemes to describe information resources is a key part of the solution to the problem.

This paper proposes a high level conceptual model for metadata for multimedia information in terms of classes of metadata, the roles of the different actors involved and the actions performed by each role. It further presents a high-le vel lifecycle model which applies to both metadata and to the information resources to which metadata refers.

Metadata is associated with an 'object', the object is an information resource. The object itself may or may not be digital (for example the documents which a library catalogue refers to may be electronic or physical documents).

Metadata can be used for management purposes. It can be used in relation to an information resource to give information about the content in order to facilitate discovery, identification, location and legitimate usage, and to track leg itimate usage (and to detect and/or prevent unauthorised usage), of that content.

We use an alternative view which treats the information resource I and the metadata M associated with it as disjoint.

This relation may be realised implicitly (eg when metadata is embedded in or wrapped around the information resource) or explicitly (eg when the metadata contains a reference or pointer to be used for accessing the resource, for exampl e a URL).

Figure 1 shows two levels of metadata, where I¹ represents information resource 1, and M¹ the metadata associated with it. Metadata M¹ is also an information resource I2, which has M² (meta-metadata) associated with it.

The model is based on three concepts:

• Roles.
• Metadata classes (attributes and their grouping).
• Actions

This model can be visualised simply as shown in Figure 2.

Three major roles are identified in the metadata model:

• Creator
• Service Provider
• User

At a certain level of abstraction these roles can be applied both to information resources and to metadata.

The user (of an information resource or of metadata) makes use of that information resource or metadata during the active life of the information resource or metadata. The user may be a human end user or an organisation, or the use may be a software agent or application.

The service provider offers services to users relating to the use of information resources and/or metadata, and making use of metadata.

Metadata structures can be arbitrarily simple or complex and metadata information can be classified according to different criteria. A useful classification of metadata elements into different groups comes from the IEEE Learning Object Metadata Working Group. This classification identifies nine classes of metadata. To these nine we add a tenth category: Security.

• General: Basic reference to the resource. This class includes an identifier and titles for the resource.
• Characteristics: Groups non-contextualized features of the resource, i.e. characteristics that are independent of its use, like language (of the user interface or content), descriptions in several languages, the subject presented, etc.
• Life Cycle: Characteristics related to the different phases of the resource. This class includes versioning information, lists the authors and the publisher, and may include information about when the resource ceases to be accessib le.
• Meta-metadata: Characteristics of the description rather than the resource, such as who created the metadata, under whose authority, using what kind of structure, who the metadata was validated by, etc.
• Technical: Technical features of the resource. This class includes data elements on format, size, location, hardware and software requirements, etc.
• Use dependent: Features which need to be interpreted according to the use of the resource. An example of such a class groups data elements that are specific to resources being used in an educational context, in which case it includ es prerequisites, objectives, target audience, difficulty, pedagogical duration, etc.
• Rights Management: Includes data elements on how the resource can be procured, whether or not support is available, its pricing, etc.
• Relation: Characteristics of the resource in relationship to other resources, like other versions of a resource, resources referred to, etc.
• Annotation: These are comments on the use of the resource.
• Security: metatdata concerning digital signature, watermarking and other security mechanisms.

A lifecycle model for information resources is shown in Figure 3. The timeline runs from top to bottom of the diagram.

In the first phase the information resource comes into being - either by creation or acquisition. Creation might involve for example the activity of authoring a document, or painting of a

In this phase the information resource is delivered for its active life in the following (use) phase. For example content is put onto media eg CD-ROM or loaded onto a server for distribution /access.

In this phase the resource is available for access and use by users. This is the main phase for which all other phases are either preparation or follow up.

Following the phase of active use of a resource, the resource may be left in place (allowed to persist) or it may be discarded - in effect deleted - (eg a website may be removed from a server, an out-of-date report may be shredded) .

The time relations between the two timelines (for information and its associated metadata admit a number of possibilities. For example the creation of the metadata may precede the existence of the resource to which it refers. At the ot her end of the lifecycle the metadata may well outlive the resource to which it refers. So the two timelines are only loosely coupled. An alternative representation of the top level lifecycle view is shown in Figure 4.