PREMIS (Preservation Metadata, Data Dictionary Maintenance Activity)
Official Web Site  

PREMIS Implementation Registry

SPAR - Scalable Preservation and Archiving Repository

Activity Name: SPAR - Scalable Preservation and Archiving Repository
Organization: Bibliothèque nationale de France
Organization Type: Library
Content Type: Audio - yes
Cartographic Material
Datasets - yes
Images - yes
Text-based Materials - yes
Video - yes
Origin of Content (digitized, born digital,
or both):
Digitized and born digital
Description: Ingesting the corresponding information packages with a METS manifest. PREMIS object and event are used inside this METS wrapper.
Start Date: 2010-05-01
Stakeholder/Audience Direct users: digital curators and content producers ; indirect users: library users, on site and on line.
Notes: General Information: System global requirements: http://bibnum/spar/index.html Notes: Active: Digitized books (printed monographs, periodicals, manuscripts), digitized images, digitized audio and video; third-party bit-level preservation. Under development: Web archives Planned:born-digital audiovisual legal deposit (including video games); records management administrative production (including records management and institutional databases).
Implementation Details:
What repository workflows or functions does your PREMIS Metadata support? Metadata storage, Digital content management, data management (PREMIS wrapped in METS and mapped to RDF for data management purposes). PREMIS is used to track all the events that occur inside the repository, not only those described in the manifest for each information package but also those tracked in our monitoring databases.
What PREMIS data module entities are represented in your implementation? Object and Event. Agent: only implemented as linkingAgentIdentifier of Event.
How is preservation metadata stored in your repository? Each information package has its own preservation metadata stored within it in the form of a METS XML manifest. All this information is indexed in a RDF triple store, with an OWL ontology per OAIS information type (representation, reference, context, structure, provenance, fixity), and an additional agent ontology, inspired by PREMIS.
What ancillary resources and tools do you use to support your PREMIS implementation? Controlled Vocabularies Locally created vocabularies.Some of them are directly expressed in our ontologies as classes (implementation of eventType) or as properties(implementation of linkingAgentRole, relationshipTypes and subTypes between objects) Tools The overall software supporting our implementation is SPAR, BnF's own modular software implementing the core OAIS functions. SPAR is a framework integrating open source bricks such as JHOVE and MediaInfo for file characterization and property extraction. JHOVE2 will be added in the next iteration. Extension Schemas Descriptive metadata, metadata about a source physical object: Qualified Dublin Core; Technical metadata: textMD for text, MIX for images, MPEG-7 for audio and video, containerMD for ARC files. Registries SPAR has its own local digital format registry. For each standard format, a link to the corresponding PRONOM identifier (PUID) is provided. This format registry also stores local profiles of standard formats. It uses RDF technology, which could facilitate deeper integration with UDFR, when it is out.
Is your preservation metadata created internally, imported from external sources, or both? Descriptive information and events that occur before the information packages are ingested into the repository. They are imported and mapped from external resources (catalogues, metadata generated during digitization, etc.). SPAR also stores preservation metadata created during the lifecycle of the information packages within the repository.
Is your preservation metadata used only internally or is it shared with external organizations? Only used internally. The publication of our reference repository (a registry containing formats and software information, repository policies and the ontologies used in SPAR) is currently taken into consideration, but PREMIS is not involved here since we intend to expose the data as RDF. Should we decide to share our reference repository, the format used would be RDF datasets.
Example File Name: 42_BnF_bc6p03r4h.manifest.version0.release0.xml
Description of Example Files and/or Additional Example Files: 42_BnF_bc6p03s26.manifest.version0.release0.xml 42_BnF_bc6p03t58.manifest.version0.release0.xml 42_BnF_bpt6k206840w.manifest.version0.release0.xm 42_BnF_bpt6k5444439j.manifest.version0.release0.xml 42_BnF_btv1b59670348.manifest.version0.release0.xml