A registry framework and Rosetta attributes for distributed science

摘要

Information is shared within organizations and between organizations at an ever increasing rate. One system design that enables sharing in a distributed environment is a Virtual Observatory. A Virtual Observatory is built with clearly specified and standardized components. One essential component is a data model for the exchange of metadata. Well designed data models are compatible with the tenets of the ISO-11179 and ISO-20943 standards for metadata registries. Another essential component is the specification for persistent universal identifiers which are used to reference resources provided through the system. Identifiers are used both to link harvested information back to the original resource description and by services to retrieve a resource. Harvested information is maintained in a registry. To harvest information from a set of metadata it is necessary to understand the schema of the metadata. We propose an approach to the formation of universal identifiers which enables one to clearly discern the schema of the source metadata. A primary registry containing a broad set of base information exists in every Virtual Observatory. Specialized Task Oriented Registries (TOR) are typically derived from the primary registry. The schema in a TOR is often flat and items in the TOR can be managed in a similar manner to that prescribed in ISO-11179. The need for interoperability within a data environment and between data environments leads to the identification of broad-based, common attributes we call “Rosetta Attributes” which will enable general cross domain searches. Access to metadata is the final component of a Virtual Observatory. Access is provided by a registry service. A specification for a minimal registry service is presented.