MediaMosa for Managing Video Content Event: The Drupal Government Days, Brussels, 7-9 april 2011 Presentation by Michel van de Ven

1. MediaMosa
for Managing Video Content
Michel van de Ven
mediamosa.org
CTO Madcap BV

2. Where are your assets?
“Help me out here. If the stuff is worth keeping,
it's worth keeping so that it's easily accessible,
right? And if it's not, then why are you keeping it?”
-- David Allen, Getting Things Done
If something is worth keeping,
it's worth knowing where it is, and what it is.
-- me
“Where is it?” -- Alexander the Great

3. Overview
● MediaMosa 101
● Features, architecture, community & Open Source
● MediaMosa: where's the value for you?

4. MediaMosa 101
● MediaMosa is built with Drupal
● MediaMosa is a Drupal Video Distribution
● MediaMosa is a
● DAM: Digital Asset Management System
● MAM: Media Asset Management System
● MediaMosa is a back-end
● For video-websites = front-ends
● For document management systems
● Generic distributed database with metadata

5. MediaMosa Features & Services
● Storage and retrieval of any digital asset
● Described with standard metadata formats:
Dublin Core, Qualified DC, IEEE/LOM, CZP
● Structured into assets, mediafiles, collections
● Authentication & authorisation (ACL)
● domains: e.g. mediamosa.org
● realms: e.g. ASelect, Entree, DigID, Federated.
● users: e.g. email michel@mediamosa.org
● groups of domains / realms / users

6. MediaMosa Features & Services
● Search using Common Query Language
CQL, SRU/SRW
http://www.loc.gov/standards/sru/specs/cql.html
● Apache SOLR used for full-text searching
http://lucene.apache.org/solr
● Support for Open Archives Initiative -
Protocol for Metadata Harvesting (OAI-PMH)
http://www.openarchives.org
● provider available, harvester in roadmap (NIBG)

7. MediaMosa Features & Services
● Transcoding of almost any video/audio format
using FFmpeg http://www.ffmpeg.org.
● Use of predefined transcoding profiles.
● Support for literally hundreds of formats.
● Distributed Job Processing Architecture
● Uploads: POST, PUT, FTP
● Notification infrastructure
● Logging capabilities
● Statistics gathering and reporting

8. MediaMosa Features & Services
● Streaming solution is not part of MediaMosa
● This is by design: every institution has its own
requirements for this, or an existing solution.
● Out-of-the box HTTP chunked streaming is
supported, so it “just works”
● DBUS challenge-response protocol used for
authenticating front-ends
● Unit Tests across the board
● Full docs available on http://mediamosa.org/api
● MediaMosa Construction Kit for Drupal

9. Operational
● Runs on standard LAMP stack: Linux, Apache,
MySQL, PHP.
But also FreeBSD, OpenBSD, nginx, lighttpd,
mariadb, varnish, memcached, mongodb, &c
● Apply existing knowledge in these fields for
scaling up & out, and to achieve redundancy,
availability &c.
● In production at Surfnet / Kennisnet since 2008
● Scales up from a laptop with a USB memory
stick, to anything.

10. Architectural Principles
● Open Source
● Open Standards
● Open Protocols
● Drupal used as a Web Application Framework
● Drupal used to build front-ends
● But... any front-end technology is possible:
Java, .NET, Blackboard, Ruby on Rails, node.js,
Android, Apple, &c: as long as it can talk REST.

11. Architectural Principles
● Service Oriented Architecture (SOA)
● Representational State Transfer (REST)
inside and outside (!)
● UNIX Design Principles
http://www.faqs.org/docs/artu/ch01s06.html
● Worse is better
http://www.jwz.org/doc/worse-is-better.html
● Modifiable & Self Documenting:
"Use The Source, Luke."

12. Open Source
● Released as GPL2 Open Source in June 2009
● Community on http://mediamosa.org
source code, releases, fora, issue trackers,
mailing lists, whitepapers, demos, research.

13. Open Source
● Product Management Committee (PMC)
● Current members are:
Surfnet, Kennisnet, National Institute of Sound
and Vision (NIBG), University of Amsterdam,
Madcap, OneShoe.
● PMC decides on roadmap, events, funding.
● MediaMosa Foundation to be created this year.

14. Open Source
● Currently using a Cathedral model
● Transfer to Bazaar model in two years

15. Users, builders & thinkers
● Archipel: chosen as central DAM after
extensive research. (e.g. not Fedora
Commons)
● Surfnet: surfmedia.nl & research platform
● Uni Twente, Uni Amsterdam, Uni Groningen,
KU Leuven, Uni Gent.
● NIBG: Ed*It platform, ed-it.nu
● Kennisnet: teleblik.nl, leraar24.nl, e-tv.nl, s-tv.nl

16. Users, builders & thinkers
● Cineca: linking 48 Italian universities with
lecture recording system (Matteo Bertazzo)
● University of Berkely chose not to use
MediaMosa for Opencast / Matterhorn, but still
uses it as an architectural reference.
● accessibility.nl / webrichtlijnen.nl / Bartimeus:
currently looking for funding to expand
transcription capabilities

17. Roadmap
● Transcription: transform speech into text
● Useful for making video archives searchable
● Making meeting recordings searchable
● In progress: “SURFconext” (TERENA/NREN),
Open Social meets Federated Identity
Management.
● Cloud functionality: Open API support &
building a cloud of MediaMosa instances
● MediaMosa as a DAM for OpenCast setups

18. MediaMosa: where's the value for you?

19. MediaMosa for Government
● Centralisation
● Of digital (media) assets
● Of access control (target groups, embargos)
● Of quality control (HD, ads)
● Single back-end for all front-ends
● shared content on intranet, internet, specific
mission sites, microsites &c

20. MediaMosa for Government
● Keep Metadata on all assets
● Know where it is
● Know what it is
● Make it accessible: online, mobile
● Make it findable
● Keep it that way for the forseeable future

21. MediaMosa for Government
● Transcription & Subtitles
● “Good enough” transcription, ~80% recognition
● Makes any video accessible via text searching
● Excellent basis for further manual processing
● “Good” transcription, ~95% recognition
● Used for valuable content
● Uses additional knowledge about speaker's voice,
auditorium acoustics, external noise like traffic,
sirens, and slamming doors.

22. MediaMosa for Government
● New results from Oslo University
● Search for specific events in video
● Uses a combination of video images, speech,
OCR, actual newspaper texts.
● Possible queries:
● “President X meeting minister Y”
● "Governor Z talking about energy plan"
● "Feynmann on General Relativity"
● “Rode Duivels schot op de lat”

23. MediaMosa for Government
● Increase accessibility of digital content by
postprocessing of image (quality, OCR, event
recognition) and sound (quality, subtitles)
● Dedicated workflows can be added to support
repetitive scenarios.
● Increase accessibility by adding subtitles
● http://www.toegankelijkevideo.nl/
● http://www.bartimeus.nl/
● http://accessibility.nl/

24. MediaMosa for Government
● Open data
● Open Social API support for mash-ups.
● Open media formats: only use open spec
containers en codecs.
● Integrating with existing federated
authentication and authorisation platforms.
● Mobile support for video: formats, streaming &c

25. MediaMosa for Government
● HTML5 / CSS3 support
● Stand-alone video players
● Semi-open players available, fully open source
player is under construction

26. Round Up
● MediaMosa
● Digital Asset Management System
● Capable, available, scalable, open, standardised,
searchable, authenticated, centralised, buzzword
compliant, Drupal.

27. Q&A
Michel van de Ven
mediamosa.org