Technical Approaches

This section is in preliminary draft form, and is publicly available with the understanding that it is not an official representation of the MMI position. This preamble will be removed once the document has been internally reviewed.   Ed.—2008.09.25

What technical approaches are available to track the necessary information for satisfying community issues?  There are several fundamental capabilities that are required: metadata collection and management, automated metadata generation and tracking, embedding metadata in served ontologies, and presenting metadata on the web.

By defining best practices, even if they are not yet implemented by many (or any) organizations, we can encourage the rapid development of more mature and community-centric repositories.

Metadata Collection and Management

Metadata must be collected at the appropriate moments in the vocabulary's life cycle. For vocabularies created and managed elsewhere, this typically means when the vocabulary is first submitted (registered) in the ontology registry. It is important to collect from the submitter everything he or she (or it, for automated submissions) knows about the original of the ontology.

Once collected, the information must be maintained (persisted) by the registry, so information does not have to be re-entered. It must be manageable by the appropriate entities (the registry manager, and/or the vocabulary originator) with easily-used processes.

Automated Metadata Generation and Tracking

A second opportunity to collect metadata is whenever a source vocabulary is updated in the registry. A number of automated metadata items, including particular the time of update, provide critical context for users of the ontology.

Metadata about the ontology can also be automatically generated by the ontology registry/repository, from two sources: the source vocabulary or ontology itself (and the derived ontology in the repository), and the transaction information maintained by the system serving the ontologies. All transactions of an ontology, from searches and views to mappings and updates, provide a rich metadata legacy that can be used in the future.

Many user roles, from ontology users to community ontology developers and semantic analysts, have a strong interest in access to the maximum possible amount of metadata about the ontologies. The recommendation to the ontology provider is to be as complete as possible when designing their automated metadata generation capabilities.

Embedded Metadata in Served Ontologies

To the extent metadata captures a stable, publicly representable fact about an ontology, it should be included as a set of elements in the served ontology. Other ontology engines can use this metadata in many ways, and can re-present it to other users of the ontology, maintaining a provenance chain back to the original provider.

OWL and RDF include standard ways to represent metadata within the file format, and this should be adopted by the ontology provider. At the application level, the Ontology Metadata Vocabulary project has produced a standard model of metadata that is important to include in every ontology.

For community vocabularies, MMI is recommending a more complete set of metadata related to community vocabulary representation issues discussed in this text. MMI presents an initial version of it on the MMI web site.

Presentation of Metadata on the Web

Finally, as mentioned earlier, it is important to visually present a representation of the source of an ontology whenever the ontology, or a significant item from it, is visually presented. The source must be presented professionally and attractively on the page, with appropriate highlighting to give a high level of recognition.