L2L
Live to e-Learning
Matteo Bertazzo - CINECA InterUniversity Consortium
3rd TF-Media Task Force meeting, Trondheim, October 2010
Bridging physical and virtual classrooms using an integrated
lecture capture and delivery service
L2L
the L2L service performs asemi-automatic transformation of live
lectures into e-learning activities ready for the publication and delivery through an e-
Learning platform
Main objectives• Productivity
“e-learning lessons-hours” / “man-hours to produce them”
• Lesson’s “Time-To-Market”low latency for publication
• Modularity of the service and investments
let’s start now with a flexible service and a plug-in model
• Content reuse by multiple servicesa L2L lesson (or part of it) delivered or used by WebTV or DTT,
web portals, other services
The context
Cineca is a non profit Consortium, made up of 43 Italian universities, the The National Institute of Oceanography and Experimental Geophysics -OGS, the CNR (National Research Council), and
the Ministry of Education, University and Research (MIUR)
founded in 1967 350+ employees acts like a trait-d'union between
the academic world, the sphere of pure research and the world of industry and Public Administration
supercomputing center, currently #70 in top500.org list (currently SP6 and BG/P, BG/Q in 2012)
References
Ministries of University and Research, Foreign Affairs, Justice Health
Universities Italian National Drug
Agency (AIFA) National Health Institute
(ISS)
Ferrari Luna Rossa (America’s Cup
challenger) ENI (Italian National Oil &
Gas Company) ESA (European Space
Agency) British Library ...
ICT & HPC services for Member & non-member
universities MIUR - Ministry of Education,
University and Research Public research institutions
Technology Transfer to: P.A. & L.A. private companies E.U.
Cineca is currently involved in a number of EU Research Projects in many fields: supercomputing, economic, physics, bioengeneering, pharmaceutical, cultural heritage…
Activities Design & management of computing
infrastructure High Performance Computing &
Scientific Visualization Data Center outsourcing e-Learning and multimedia services
LMS (Moodle, BB): integration, management, customization of large-scale platforms (ASP model)
3D Virtual environment based on SL/Opensimulator
EU Projects e-Learning consulting Streaming services and now… Lecture recording
and delivery services Information System design,
development & analysis Network management Multimedia Communications DataWareHouse Services Portals design and development Digital Libraries
Main Activities &
Resources
Resources HPC Systems: High Performance
Computing platforms for technical and scientific computing
IT Systems: "Mission Critical" platforms used to host the IT services offered by CINECA
Data Storage environment: a high-end Storage Area Network connected via fiber channel and serving all the computing systems
Graphics and Virtual Reality: a collection of instruments and expertise for scientific visualization
Multi videoconference and Access Grid (AGnode): point of access to advanced systems of multivideoconference
Specific inputs/reqs vs.
consortium approach
L2L service designed to match specific requirements and the consortium nature: rationalizing resources putting in place “economies of scale” supporting different service models
• providing good services containing costs• Currently reqs are coming from 5 universities (2
distance learning univs.)
L2L service
components
DAM
Player
Recorder
Recorder
Recorder
Modules
I
N
T
E
G
R
A
T
E
D
D
I
Y
--
R
E
M
O
T
E
O
F
F
-
L
I
N
E
MEDIAMOSA – MOODLE
INTEGRATION
STORE - MNGMT
Unique centralized
DAM
(CINECA ASP)
Multiple Univs
Multiple Moodles
(CINECA ASP or
Univs in-house)
DEPOSIT
UPLOAD
L2L RECORDER
RECORDING
L2L PLAYER
DELIVERY
On-line|Off-line
Different usage scenarios
Modules UNIV-B(In-house)
UNIV-A(CINECA)
BB
HTML5 PLY
(not yet)
Adobe Flash Media
Player
Multiformat delivery and
podcasting
Mosaic
Podcasting
L2L –
Architecture
Designed following the SOA model used by MM REST interactions
L2LRecorder MD
Streaming Servers
QT
WM
RN
FM
Moodle LMS
MoodleBlock
Moodle Module
L2L Player
Other Services
Concept Mapper
Speech-to-text
Transcoding Server
Mediamosa DAM
L2L Ingester module
FS
Still Server
Download Server
WebDavFTP
Upload
HTTPRest
HTTP Rest
HTTPRest
NFS
Streamingprotocols
Ingester module in order to support L2L lessons’ structure unpack and ingest L2L packages coming form stations plug-in additional services (internal | external services using)
L2L –
recording station
L2L station: we provide our univs. with HW and SW L2L Full: designed to be integrated in a lecture hall – classroom with
A/V facilities (PC, video capture card, PTZ Camera, VGA2USB adapter, Mic, secondary 7” display) + L2L Recorder SW
L2L Light: laptop (i5+ class CPU) + Mic + WebCam + L2L Recorder SW
Functionalities Performs the Recording (up to 2 video sources) Supports the recording description
automatically: retrieves user profile, Moodle platforms, Moodlecourses using an optional SAML auth (Shib) and WSs;
manually: title, description, tags Performs Sync and MD/Content extraction from MS PPT
presentations (titles, text, JPEGs for thumbs, slide change events) Scheduled/On-demand upload to the DAM using FTP/WebDAV
Features 5 recording modes (A, A+Scr, AV, AV+Scr, AV+Ext) Supports off-line recording HQ (1Mbps each video) recording in MS WM format XML descriptors (technical MD, descriptive MD, structural MD, sync
MD) Content packaging (custom)
L2L Recorder -
workflow
Manual/scheduled upload
Quality control
Recording (sync)
Attachments selection
Lesson type selection
Manual description
Automatic metadata (WS)
Authentication (SAML)O
P
T
I
O
N
A
L
DAM -
Mediamosa
the well-known Open Source Digital Asset Management Platform by SurfNet…
production env. based on a clusters installation of the 1.7 version
MM Functionalities in L2L: Mngmt: assets creation, MD mngmt,
collections (faculties),… Analysis/Transcodig: mediafiles
analysis, transcoding profiles for FLV-x264-mp3/Ogg-theora-vorbis/MP4-x264-aac conversions
Multi-format delivery services: web publication (emb., pod.) and streaming (existing streaming farm for QT, WM, Flash and RM)
Storing all lessons’ files (videos, XML, JPEGs, Attachemnts, PPTs, …) in MM
L2L lessons’ representation: extended DC MD schema in order to
define relationships between MM assets: L2L lesson = main asset (containing 1 main XML descriptor) and related assets (containing all other files) sharing commons set of descriptive MD
MainAsset
Asset Asset Asset
ConceptsXML
FLV transc.Mediafile
WM Original Mediafile
L2L Lesson
L2L Package
IngestionModule
L2L Recorder
Asset
XML Desc Mediafile
Asset
JPEG Slide
L2L and DAM –
functional schema
Streaming
IngestPublish
Manage
Store
Mediafiles Mediafiles Mediafiles Mediafiles
Assets Assets Assets Assets
Analyze
Transcode
Metatag
Still
Streaming
Download
WebPub
View Access
Workflow Metadata
Producers Editors Audience
L2L Portals WebApps Modules Portals e-LearningWebTVUploads
External services
L2L –
Moodle module
A standard Moodle module (v1.9+) L2L activity usable among other
activities inside a course Following the Moodle model, the same
module for mngmt and viewing Fully compliant (backup, restore)
Functionalities: browsing/searching/selecting lessons
using structural and descriptive metadata
Managing visibility of content (presentation, attachments) in the course context DEMO
L2L Player visualization Currently flash based Getting all the content from
Mediamosa (Mediafiles, JPEGs, Attachments, Presentations, XML descriptors)
A/V visualization and sync Lesson “navigation” by:
Direct access to the timeline Slides thumbs Table of contents (slides’ titles) Full-text search results
Concepts DEMO
Teacher view
Student view
Providing studentes with the ability to browse the lesson using a list of concepts pointing them to the specific slide where the concept is being discussed
L2L integrates ConceptMapper using a REST-API, a tool developed by Cineca in the context of the European project Papyrus (news items’ annotation)
ConceptMapper (built on top of WikipediaMiner) exploits Wikipedia information to automatically identify and disambiguate relevant concepts in texts (it, en, de, fr).
outputs consists in a list of concepts identified by the titles of the Wikipedia pages that describe them
For each concept a relatedness measure to the document or to a domain is provided
L2L applies two filters:
a threshold on the relatedness measure
selecting concepts of a slide only if they are also identified on the entire presentation
Concetta –
Concept browsing
http://www.ict-papyrus.eu
Sync (SLide)Concepts
SlideText
Publish
Concetta –
workflow
ConceptMapper
DAM
SlideText
SlideTextSlide
Text
L2L Package
IngestionModule
L2L Recorder
Concepts
• SCORM-compliant player (mandatory for distance learning universities)
• L2L Mosaic
Automated post-processing based on the GStreamer Framework
Getting an unique A/V file from a L2L lesson according to the lesson type and a skin
DEMO
• New HTML5 – open video player
• Web-editing of lessons
Current works/developments
Future works/developments
Mediamosa 2 (Drupal 7) and Moodle 2 Evaluate the opportunity to integrate new lecture recording software (Matterhorn) L2L lessons inheriting MD from Moodle courses where they are actually used Joining “Lecture Capturing” & “Digital Library” worlds
digital library software integration (Fedora Commons) adoption of standard/specifications and models for born digital content preservation
(i.e. METS for import/export ?, OAIS model for preservation) Definition of preservation policies
Integration of Speech-to-text services Currently the Ingestion module
extracts the HQ audio from videos Testing IBM VoiceTaylor (was IBM
Attila research prj) and PerVoice 2 levels: concepts identification
and subtitling for accessibility (manual correction)
Use an external service and develop language models for specific domains (i.e. medicine, law, history)
New techniques for A/V segmentation (currently L2L relies on PPT)
ePub delivery (does it make sense in this context? Maybe for some scientific domain) DEMO
“Preserving digital content for
future generations”
not only backups and disaster recovery plans
SyncConcepts
Sync TextPublish
speech2text, concepts,
subtitling
Audio
Corrections
Sync Text
« Lorem ipsum dolor sit amet, consectetur adipisici elit,
sed eiusmod tempor incidunt ut labore et dolore magna aliqua.
Ut enim ad minim veniam, quis nostrud exercitation ullamco
laboris nisi ut aliquid ex ea commodi consequat.
Quis aute iure reprehenderit in voluptate velit esse cillum
dolore eu fugiat nulla pariatur. Excepteur sint obcaecat
cupiditat non proident, sunt in culpa qui officia deserunt
mollit anim id est laborum. »
ConceptMapper
DAM
Speech2TextService
Cineca
Languagemodels
WebTV
Back-office
GESTIONE
Servizi
Di
E-learning
Modulo L2L
Docente -
Gestione Lezione
StudentiDigital Media
Asset
Management
Service
PUBBLICAZIONE
WebTVDelivery Storage
TranscodingMetadata
Applicazione
Portale
L2L
Stazione di
registrazione
OAI-PMH
DAM
integration scenarios
Thanks!
Matteo Bertazzo - IKMS Department
CINECA – InterUniversity Consortium
Via Magnanelli 6/340033 Casalecchio di Reno (Bologna), Italyhttp://www.cineca.it