RDF Representing filesystems in RDF (anarchivist)

From CURATEcamp
Revision as of 19:59, 16 August 2011 by 68.65.169.144 (talk) (Created page with "rdf for file systems Mark Matienzo, digital librarian at Yale use case creating or recieving assets -- discs don't have time to make assessment of content disk imaging with fil...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

rdf for file systems

Mark Matienzo, digital librarian at Yale

use case creating or recieving assets -- discs don't have time to make assessment of content disk imaging with filewalk which captures structure

should we use RDF? collect same data, but going forward, could assign identifiers. Does a symbolic link have to have a separate representation, or simply represent it in the metadata?

could define relationships between assets. For example, may find multiple assets which are part of a single record.

Don't have right middleware to tie this together.

Any ontologies/ vocab that people think are useful

Use would be internal? Yes. Metadata for management of the content. could be translated at some point into archival descriptions.

but a file is not always a record -- a record can be an compendium of files.


identifyng restrictions. intellecuatl content as opposed to physical content.

Automated?

Vulca-extractor??? to select content

if represented in some form of structured metadata, we could implement restrictions until we can get back to it.

Unified digital file registry is based on RDF

No complex representation of container

could use UDFR and make representation that say this representation is in this fo9rmat

planning to expand UDFR

Relationships to gather automatically:

Provenance (who owned the file), when was it created? file names the same?, info about file system

RDF would provide a finer granularity about files or directories or even disk images.

DFXML was designed as a shallow XML structure expected for incorporation into other models.

Take output and translate it

Could this help with arrangement and description issues?

Peter Van Garderen: What are the use cases? What problems are we trying to solve?

JHOVE2 will analyze file directory, transverse it and identify fies, create a virtual file type for directory and its content

Plugins? RDF is using filename pattern matching

XML is not object-oriented. Massive METS files are not the answer. Difficult to do anything with those complex files.

Core issue, problematic in PREmis. So many ways to implement it


Is the value of METS the schema? Can we replace METS with RDF?

Is there an RDF version of METS? (no)

Do we need a constrained ontology for RDF to avoid creating the same problem?

People use METS so differently.

Don't we need to agree on how to use RDF, so we don't recreate the same problem?

Still, even if someone does it a little differently, it won't break it. In METS, a tiny difference has drastic consequences.

METS implementation of relationships is site-specific, difficult to implement, hard work, difficult to parse.


Need application that can parse METS or other metadata, pull in linked data and store it with extractions in assertions, and store those.

Mark Matienzo: EAD and METS should be serializations. We should build them dynamically from something else.

UCSD schema for METS on LOC website is one of the best.

They store as RDF and export into METS for transmission.

That's how it should be used -- for transmission. Some store it in METS but then have to version it.


More flexible to store information and export it in the version/format you need.


if RDF enables content to be more usable, then it's an improvement. How it is used externally is most important.


could potentially provide new avenue for discovery