Metadata Collections & QA


Concept of metadata interoperability

The actions embodied by metadata interoperability are best explained in the article, Combining RDF and XML Schemas to Enhance Interoperability Between Metadata Application Profiles. These actions include the ability to:

  • apply a single query syntax over descriptions expressed in multiple descriptive formats
  • express the relationship between multiple descriptions in terms of a "core" or "canonical" description
  • project community or individual specific descriptions out of a single canonical description

This means that collections which support interoperable metadata protocols allow remote clients to search and retrieve their metadata records. These remote clients can then provide their users with search results from another source, such as DLESE.


Forms of metadata interoperability

The actions described above lead to two forms of metadata interoperability:

  • Federation: Service requests are handled through several service providers that distribute queries to remote metadata providers, collect the results, and publish these results in response to the request.
  • Harvesting: Service requests are handled through a single service provider. The service provider collects (i.e., harvests) metadata from metadata providers. Service providers then import these metadata into their discovery systems.

Federation requires more effort from data providers but is easier for service providers (i.e., service providers do not need to parse and interpert metadata records). It has, however, scalability issues (i.e., it rapidly becomes unreasonable to distribute a query to more than a few sources). Discovery facilitated through harvesting requires less effort from data providers, but is more difficult for service providers (i.e., the service providers are delivered full metadata records; they then must do something with them for these records to be discoverable in their systems).


Metadata requirements for metadata interoperability

No matter which form of metadata interoperability is pursued, the following information about a metadata framework is required:

  • Semantics: The meaning of elements
  • Structure: Human readable format
  • Structure: Machine readable format (e.g. XML)
  • Syntax: Grammar used to convey meaning and structure

Remember that metadata is information about objects (i.e. title, description, author) while data refers to the digital objects and their representations.


DLESE's approach to metadata interoperability

DLESE is using the Open Archive Initiative-Protocol for Metadata Harvesting (OAI-PMH) to provide and harvest metadata records. For more technical information on this protocol please see either website below:

As a brief summary, the OAI-PMH defines two classes of participation:

  • Data Providers: Participants who use OAI-OMH to expose metadata about the content they hold
  • Service Providers: Participants who use OAI-PMH to collect (harvest) metadata from Data Providers and use the metadata for value-added service


The underlying technology involves:

  • HTTP requests and XML responses
  • XML, XML namespaces, and XML schemas

The protocol policies require:

  • unique identifiers and guarantees of persistence
  • an OAI metadata record
  • support for multiple metadata formats
  • datestamps [the date of creation, deletion, or latest date of modification of an item, the effect of which is a change in the metadata of a record disseminated from that item (archive)]
  • flow control mechanisms

The description of the harvesting protocol includes request/response types, with the following six request types defined:

  • Identity: Self description of an Open Archive
  • ListMetadataFormats: Metadata formats supported by the archive
  • ListSets: Addressable subdivisions contained within an archive
  • GetRecord: Metadata for a single record
  • ListRecord: Metadata for all records within specified constraints
  • ListIdentifiers: Identifiers for all records within specified constraints

In order to be a Data Provider, an existing archive must:

  • maintain add/modify/delete datestamps for records
  • keep track of deleted records
  • have unique identifiers for records
  • supply metadata in unqualified Dublin Core format and optionally, other metadata standards
  • technology to implement the OAI protocol

A list of archive policy decisions is useful also.

The OAI metadata record has the following sections:

  • Header: Information common to all records (identifier and datestamp)
  • Metadata: A single manifestation of the metadata from an item in an unqualified Dublin Core format
  • About: Optional section about the metadata part of the record (right and usage information about the metadata part of a record)


Last updated: 5-22-02