Modern Repo Brainstorming Activity

What does a Modern Repository look like? (Forget all about DSpace/Fedora/EPrints/etc. and just describe how a "repository" should fit into the modern web. What do users expect out of a "repository"? What features should a Modern Repository have? What user needs/wants should it meet?)

• not just research anymore: photos, music, data sets, video, etc
  • More different kinds of content and metadata schemas
• research management systems
  • CRIS moves the repository to the back-end. As CRIS will be the front end
  • In edinburgh, PURE is being used with the LNI to ingest
  • Needs to support "whole research workflow", Grants, etc.
• simple (visual?) import – think dropbox?
  • Excel, Drag & Drop, minimal work to get things in
  • DepositMO
  • SWORD / SWORD2
  • ScottP: submission should be much much easier.
  • Bram: ScribD also had very easy upload, but poor in metadata. Nice feature in embedding lists & collections in other applications
• automated metadata capture
• content easy to use / reuse
• CRUD (needs to be able to easily Create, Read, Update, Delete – especially via web services, e.g. SWORD2, etc.)
• Support for branding / theming
• Support for Customizations (metadata and metadata structure)
• Various storage system integrations
• Flexible content workflows
• Support versioning / relationships
• flexible authorisation
  • give more control to user communities (do their own branding, etc)
• Support complex objects (representation of), both human- and machine-readable
• scientific data sets
• reporting (generate useful reports
• Support content reuse ("open" data)
  • e.g. embed info in dept website
• search (easy)
  • faceting / filtering (e.g. Solr)
• statistics: regular reports to item authors (like Digital Commons), plus usage/admin reporting
  • bot filtering (automatic?)
• getting stuff out
  • support disciplinary aggregation
  • creating adhoc "sets" of content
  • (this made me think of http://www.apsr.edu.au/orca/ - Kim)
• shareable metadata
  • different metadata "views"
    • shared version vs local use metadata
• new name (not "institutional repository"): maybe just "repository" or "storage"?
• support for preservation activities
• identifiers / persistance (flexible, granular, parts of items, people, collections)
  • the perils of handles...
  • DOIs vs Handles
  • Truly external IDs
• support access / privacy - making things "dark", either temporarily (embargo) or permanently
• "repository / DAM system that can display stuff vs. CMS that can do DAM"
  • do one thing, do it well
• flexible metadata schema
• dissemination
  • make data usable / able to be "mashed up"

Some Additional Details / Specifics:

• Richard Jones: DepositMO project
  • dropbox like interface
  • integrated in Microsoft Office.
• Mark Diggory
  • Changing Branding
  • Presentation of Metadata
  • How well the metadata can be structured in a repository
  • Integration with other storage systems
  • Customization on the workflows
• Richard R?
  • Versioning
  • Relationing technology: linking items and content to each other
• Sarah Shreeves
  • Giving more control to the user communities: delegated administration on steroids
  • Good representation on complex objects
• Adam Field (eprints services)
  • Reporting (getting data out of the repository in a non-publications form)
  • Integration with other websites
  • Search data well
• Bram
  • Statistics for Repository authors to get more traction
  • annual reporting gets more demanding
• Richard
  • Faceted browsing
• Brad mclean
  • Getting things in & managing
  • Getting things OUT again
  • boundaries of a repository are institutional, but ideally you might want to break down those boundaries and make them more disciplinary
• Richard rodgers question: did OAI-PMH failed in this regard?
• Mark Diggory: one problem related to that: missing official taxonomies?
• Elin: problem for disciplines that have no good aggregators yet. It would be a huge motivator for them.
• Richard Jones: Creating ad-hoc sets that can be easily shared
• Bram: the issue is not technical. Which organization will put up the effort & the money to make these big portals possible. (example economists online)
• Stuart Lewis: could not be "one" repository, but different ones. It's a service, not one monolithic thing
• Elin: focus is not about offering storage, but offering assistance with their content and metadata
• Robin: "for completeness" we can include Preservation. Justification for having the repository is that it does the preservation (versus CRIS).
• Stuart: the features for preservation are present only need to be used
• Tim: DSpace enables preservation but doesn't do it in itself. Preservation requires people, policies. No technology/system does preservation – it can only enable one to perform preservation activities.
• Jones: Person identifiers
• Diggory: all of your accounts, personal accounts in DSpace?
• Scott: I hate handles
• Rodgers: handle vs internal identifiers. Google Scholars … handle are the work of the devil (no content in the URL's). Exposing metadata in the URI is the way to go.
• Scott: main point for handle "moving content" to other insititutions is not possible in DSpace.
• Sarah: Handle "forces" a good practice in standardized URI citation.
• Stuart: Academics start to understand DOI. Crossref. Selling handles as DOI's.

• Mdiggory: Nescent work with DOI registration. True sense of the use of identifier. Identifiers lives outside of the repository.
• Jones: the modern repository needs to provide an identifier. But not sure whether it should be persistent, or be linked to an external service. Granular identifiers required (datasets, bitstreams, …)
• Mdiggory: privacy & private items, access controls.
• Jones: do we really want to add a lot of CMS features?
• Hardy: more user interface support for UI's. map viewers, image viewers, …
• Mwood: why doesn't do the browser do it?
• Robin & Richard: how do the fedora folks look at this? Fedora should be pretty good at the "invisible repository" bit.

Whiteboard photo (click for full resolution):

Additional Close-up Pictures of sections of Whiteboard:
(click for full resolution):