Ryan's notes from Open Repositories 2007.
Random thoughts
These thoughts are very random snapshots of Ryan's experience.
Leslie Johnston says: data sets are even harder than video.
Sally Jo Cunningham presented on a very informal, yet informative, study of Institutional Repository usability.
Tufts is working on a service called myDL that gives users more convenient access to the Fedora content. Details??
Several projects, including eSciDoc and RepoMMan, mentioned that users will not be allowed to delete content once it is placed in the repository.
Matthias Razum of eSciDoc presented on a system that can provide object-level versioning (rather than datastream-level). This is particularly important for objects that are stored in an atomistic content model, since a datastream change at any level can create a new version of the object. Code should be available soon...
FedoraGSearch can support multiple indexes, which may make our searching system easier to implement. GSearch also supports highlighting of matched terms in result sets. It would be worthwhile to look through this code.
MPTStore is a replacement for the Kowari triplestore, based on a simple database structure. It is very efficient for common types of queries, and doesn't have all the extra administration overhead of Kowari. There is one table per predicate. Currently, it only works with postgres.
mod_oai lets you deliver OAI records generated from websites. This allows better coverage than just relying on a search engine crawler. Many search engines will treat an OAI result set, so you get a benefit from using mod_oai even if the search engine doesn't explicitly support OAI.
Some of the MIT SIMILE tools may be useful. In particular, Longwell is an interesting front-end for faceted search. Dwell is Longwell adapted for DSpace.
Future of Fedora
The big news is that Fedora will be moving out of Virginia/Cornell, and into the Fedora Commons Foundation. For more details, see the Jester's blog entry on Sandy Payette's plenary talk
.
Interoperability
The community is starting to really focus on repository interoperability. Projects include:
- Asset Actions (as always)
- OAI-ORE
- JISC Common Repository Deposit Service Description
- National Library of Australia is developing a METS profile for repository interoperability – we should follow this.
- The CAIRO project is developing a METS strategy for storing archival objects – this may be related to Sound Directions, but it isn't very far along.
Preservation
MacKenzie Smith's presentation on preservation issues was particularly interesting.
- PLEDGE is a system for the concrete specification of preservation rules, based on METS and Rei. It allows an organization to take a preservation checklist and formalize what is meant by an "adequate number of copies" of an object, how often integrity should be checked, etc.
- There is a new system called iRODS that is supposed to be the successor to SRB.
Fedora Content Models
Virginia's disseminators directly map to end-user requirements.
RIFF has a great system for validation of Fedora objects or pre-ingest files. Code should be available soon (check slides for the address). They can validate:
- basic XML (with schemas and schematron)
- Java object relationships (using both internal Fedora structures and a new validation API)
- specific content model schemas
- RDF relationships
Two working groups were formed related to content models:
- Content Model Formalization WG
- Disseminator WG
Fedora Interfaces
At birds-of-a-feather session on Fedora interface design, it was decided that all of the participants would write descriptions of the tools they are building, and post these descriptions to the Fedora wiki. This will help to avoid duplication of effort.
