Time-Based Media Breakout
HydraConnect, San Diego
January 24, 2014

Attendees: Michael Klein, Mark Bussey, Joe Pawleko, Karen Cariani, Susan Pyzynski, Richard Green, Jim Tuttle, Claire Stewart, Mark Notess, Jon Dunn, Patricia Diaz, Katherine Lynch, Hannah Frost

Potential Discussion Topics

• HydraDAM & Avalon (& OpenVault) ((& homegrown Hydra))

• Toolsets

• Best practices

• Managing really large collections of bits

• Video file ingest

• Discovery / Accessibility

• Can Hydra Help w/Management

• Digitization & Preservation (is this a separate concern from ingest?)

• Archiving Digital Course Assets (esp. video) (How is a digital course object different than an exhibit object)

• Avalon Course reserves

• In the "UI" and/or in the "Repo" there's some tension here

• Fedora 4... are there ingest and preservation patterns

Attendee interests and use cases:

• video and audio collections: backlogs and imminent acquisitions to ingest

• mixed media collections, such as IR

• course reserves, including temporary/arbitrary collections

• supplementary media associated with publications

• student generated video (public speaking, teaching, learning objects, etc.)

• Digital Course content

• Video Preservation

Avalon is access oriented. What can Avalon learn from HydraDAM regarding the preservation environment
Is it possible to generalize the AV functionality for Hydra, develop some best practices and community patterns?

• connection to streaming server

• access control

• delivery / players

• transcoding

• preservation storage

Need to compare road maps for Avalon, Open Vault, HydraDAM, and Stanford projects (Spotlight! and Latvia collection) to see how they line up, in order to see where efficiencies and collaboration can occur? Avalon and Spotlight! and Hydromata together -- a humdinger!

Differences between HydraDAM and Avalon:

• preservation

• descriptive metadata: HD uses pbcore; Avalon uses MODS

• structural metadata: no tangible work yet. let's be sure to collaborate on this! -- capture as user stories

RDF, METS, PBCore -- need to be able to express structure in any of these.  RDF implementation would be great place to start, but this area is not settled enough and could hold up progress. 
-- interaction of access restrictions WRT to structure
-- does enduser annotations or other contributions play into structural issues?

What is different about time-based media?

• I/O intensive for derivative processing

• annotations -- proxy (frame-accurate or not) and master, sync issues

• retrieval - partial or full

• specialized server needs

• big files

• frame-level analysis

How can these solutions be broken down for broader adoption by others?

• Avalon: batch ingest is separate gem, rest of functionality is in the app as monolith

• stream player is discrete; this could be turned into a module for general use

• authorization: with a little more work, could be turned into a module

Action Items:

• Structural metadata use case gathering: Stanford, NYU, IU, NW, WGBH will do this

  • Mark Notess to share a template

• Discussion btw WGBH and IU re: HSM integration needs and plans

  • Does this require Fedora 4?

• Exploring what components can be abstracted and turned into generalizable modules

  • media player -- for streaming server or progressive download over http

  • authorization

  • what further decomposition of Avalon is needed to support adoption, collaboration with other Hydra heads/repositories (Stanford, Hydromata)

  • Stanford may be able to help with refactoring for broader adoption, reuse by others