Transcript Videoleectures.Net developments

VideoLectures.Net
developments
OCWC Learning Process
OCWC Overview – Textual courses
Tutors
Moodle
LMS
(Learning Management System)
Blackboard
EduCommons
Learners
Textual vs. Video Courses
OCWC Learning Process
Tutors
Tutors
Media Content Creation
LMS
Metadata Enhancement
(Learning Management System)
Web Dissemination System
Media Content Delivery
Learners
Learners
Video Enabled Learning
OCWC Learning Process
Tutors
- Scheduling
- Recording
- Editing
- Transcoding
- IP Rights Clearance
- Archiving, …
Media Content Creation
Metadata Enhancement
Web Dissemination System
Media Content Delivery
Learners
- Keyword extraction
- Categorization / Tagging
- Speech Recognition
- Transcription
- Subtitling
- Audio Indexing
Web and Media distribution
Community Building
Discoverability, Search
Pedagogic Processess
Accessibility
Branding
Scalable Hosting
Network Accessibility
Geographic Distribution
Bandwidth
IP / Content Rights
Video Enabled learning
OCWC Learning Process
Tutors
- Scheduling
- Recording
- Editing
- Transcoding
- IP Rights Clearance
- Archiving, …
OpenCast
- Keyword extraction
- Categorization / Tagging
- Speech Recognition
- Transcription
- Subtitling
- Audio Indexing
Semantic
Services
Web and Media distribution
Community Building
Discoverability, Search
Pedagogic Processess
Accessibility
Branding
Videolectures,
BYO...
Media Content Creation
Metadata Enhancement
Web Dissemination System
Media Content Delivery
Learners
Scalable Hosting
Network Accessibility
Geographic Distribution
Bandwidth
IP / Content Rights
OpenCDN
OpenCast Matterhorn
Automatic categorization
Recommender system
Speech to text
Categories
Interest
Automatic Categorization
• Use textual information: Title,
Abstract, Slides, Articles (extract text
from PDF, PPT, …)
• Use link information: Who is the
author, What is the context, What is
the origin, Which course, Which
event, Which Department
• Use Web Statistics (People who
clicked this also clicked…)
• GENERATE CATEGORIZATION
and TAGGING
Automatic Categorization
Recommendations
Speech Analysis
…
<p begin="0:45:31.860" dur="0:00:10.780">
we all know that it takes about four
point eight seconds we've done many
experiments in three meters ICS about
ball point eight seconds so the whole
thing will go very fast and going to put
the monkey up their
</p>
<p begin="0:45:45.580" dur="0:00:09.030">
I want you to foresee the trajectory of
the golf ball before rehabbing the
monkey and it is already so painful
forties monkey you don't want in two
</p>
<p begin="0:45:54.950" dur="0:00:02.200">
prior experience was going to happen
</p>
…
Multidisciplinary video
segmentation
Search through
speech
Transcription
support
Semantic technology wrapup
•
•
•
•
In production at Videolectures.NET
Near Availability for third party projects
SOA / Standards Based Access
HTML / Javascripts / AJAX snippets
available for simple integration into web
pages
Video Enabled learning
OCWC Learning Process
Tutors
- Scheduling
- Recording
- Editing
- Transcoding
- IP Rights Clearance
- Archiving, …
OpenCast
- Keyword extraction
- Categorization / Tagging
- Speech Recognition
- Transcription
- Subtitling
- Audio Indexing
Semantic
Services
Web and Media distribution
Community Building
Discoverability, Search
Pedagogic Processess
Accessibility
Branding
Videolectures,
BYO...
Media Content Creation
Open CDN
Metadata Enhancement
Web Dissemination System
Media Content Delivery
Learners
Scalable Hosting
Network Accessibility
Geographic Distribution
Bandwidth
IP / Content Rights
OpenCDN
Open CDN
• Solve the media content distribution
problem
• Cheaply (e.g. Sustainable)
• Scalable
• Geospatially
• Make content accessible in the
developing countries
Videolectures.NET - Planned System Structure
Page rendering
servers
Database
Servers
(PostgreSQL)
WEB page
requests
Load
balancer
Web page rendering
cloud
SQL
requests to
database
Data
Replication
(LinuxVirtualServer,
Apache/Django)
Page/Snippet/Data Cache
(Memcached)
Page
responses
Internet
Primary file servers
(Uploads / CDN Source)
Media
requests
OpenCDN
Content Distribution Network
CDN Peer
Host
CDN Peer
Host
CDN Peer
Host
Optional 3rd
party file
hosting
(Akamai,
Amazon S3)
OpenCDN
The technology…
BitTorrent
•
•
•
•
•
•
•
Don’t reinvent the wheel
Proven – running 15% of Internet traffic
Reliable and Scalable
Simple
Web Standard
Multiple existing implementations
Extensible
OpenCDN: Based on BitTorrent Network
Media
File
File
Server
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
OpenCDN: Torrent tracker provides real time
information
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
OpenCDN: Web server uses torrent tracker
to generate web pages
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
Web
Server
OpenCDN: Web server uses torrent tracker
to generate web pages
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
Web
Server
Page
Template
Multimedia
Web Page
OpenCDN: How to start distribution
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
Web
Server
OpenCDN: Torrent solution is RSS Feeds
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
Web
Streaming
Server
Web
Streaming
Server
Web
Server
OpenCDN: Different feeds for different server
capabilities
Media
File
File
Server
Torrent
Tracker
BitTorrent Network
http://
rtsp:// mms://
...
Web
Server
OpenCDN: Shared and Decentralized
Organization #1
Organization #2
Organization ...
BitTorrent Network
http://
rtsp:// mms://
...
http://
mms://
http:// rtsp:// mms://
OpenCDN: Reaching distant corners of the world
Organization
#1
Organization
#2
Organization
...
Organization
...
BitTorrent Network
http:// rtsp://mms:// ...
http://
mms://
http:// rtsp://mms://
http:// rtsp://mms://
Open CDN
• Based on P2P Network: BitTorrent
–
–
–
–
Proven solution, millions of users (stable and tested, well understood)
No Centralized Control, No Central Point of Failure
Optional video streaming - P2PTV: Miro, p2p-next…
Real-Time tracking of each file using existing client software
• BitTorrent not 100% perfect out of the box
– But the protocol was designed to be extended in the first place
– We are planning a few simple small backward compatible extensions
• Issues to solve
– Reindexing .torrent files and Checksum Calculation
• Same file with a different name ‘is not the same’
• Add specific SHA checksums for each file
– Scheduling files to the network
• RSS Feeds
– Fairness, Statistics
– Legal issues and content removal