intelligent tools-mitja-jermol-2013-bali-7 may2013
TRANSCRIPT
Advanced methods and tools for Web-Based
education Technologies for Structuring, Analyzing, Modelling, Personalisation
OCWC Global Conference 2013Bali
Organiser and Sponsors
• Knowledge 4 All Foundation, London (http://www.k4all.org/)
• FP7 project Translectures (http://www.translectures.eu/)• FP7 project MediaMixer (http://mediamixer.eu/)• FP7 project X-Like (http://www.xlike.org/)• FP7 project Pascal2 (http://pascallin2.ecs.soton.ac.uk/)
Agenda• Brief introduction to the workshop (Mitja Jermol, K4A, JSI) (15’)• Introducing K4A (Colin de La Higuera, K4A, Nantes) (15’)• Some challenges in using Machine Learning in OER (Colin de La Higuera, K4A,
Nantes) (15’)• Break• Technologies and solutions for content structuring (Marko Grobelnik, K4A, JSI) (45’)• Technologies and solutions for user modelling and analytics (Marko Grobelnik, K4A,
JSI) (30’)• Break• Technologies, solutions and prospects of OER in the Videolectures.net case (Mitja
Jermol, K4A, JSI) (45’)• Wrap up and discussion (20’)
Multimodal and
multilingual content
structuring
Learning analytics and
user modelling
Personalisation, recommendation
Technologies
Graph/Social Network Analysis (GraphGarden/SNAP, IST-World, FPIntelligence)
Complex Data Visualization (DocAtlas, NewsExplorer, SearchPoint)
Computational Linguistics (Enrycher, AnswerArt)
Social Computing/Web2.0 (LiveNetLife)
Decision Support (DEX)
Light-Weight Semantic Technologies(OntoGen, OntoBridge)
Deep Semantics & Reasoning (Cyc)
Statistical Machine Learning
Data/Web/Text/Stream-Mining (TextGarden Suite of tools)
From libraries to courses…
From courses to lectures….
From lectures to mass education…
From mass education to mass accreditation…
Videolectures.net development
WSA UNESCO AWARD
Personalisation
Modeling
Log files
Content m
ining
(Needs and preferences)
Adaptation
Towards personalisation @ videolectures.net
Enrycher(Contextualisation of
content objects)
Quintelligence Miner
(user modeling and segmentation)
Recommender(Content/user matching)
Content/learning object
User behavior
TEL environment(videolectures.net)
Transcribing and Translating Videos
April 14, 2023 17
Translectures
Our aim is to develop innovative, cost-effective solutions to produce accurate transcriptions and translations in VideoLectures, with generality across other Matterhorn-related repositories.
Three scientific and technological objectives:• Improvement of transcription and translation quality by massive
adaptation.• Improvement of transcription and translation quality by intelligent
interaction.• Integration into Matterhorn to enable real-life evaluation.
18
Results
Freely available tools and services for accurate transcriptions and translations:- Automatic transcription of videos: English, Slovenian, German, French, Spanish- Automated translation of videos: en es, en sl, en⇆ ⇆ fr and ende.
Initial implementation:- Videolectures- poliMedia- Matterhorn
19
Project factsheet• Total Cost: €4,491,143.00 • EC Contribution: €3,125,000.00 • Execution: From: 11/2011 To: 10/2014
No Name Short name
Country ExitMonth
Exitmonth
1 UNIVERSITAT POLITECNICA DE VALENCIA UPV Spain 1 36
2 XEROX SAS XEROX France 1 36
3 INSTITUT JOZEF STEFAN JSI Slovenia 1 36
3+ KNOWLEDGE FOR ALL FOUNDATION K4A UK 1 36
4 RHEINISCH-WESTFAELISCHE TECHNISCHE HOCHSCHULE AACHEN
RWTH Germany 1 36
5 EUROPEAN MEDIA LABORATORY GMBH EML Germany 1 36
6 Deluxe Digital Studios Ltd DDS UK 1 36
22
Massive Adaptation
massivetrain from
large and diverse data collections
adaptation target test domain
Overview About Adaptation
23
acoustic modeladaptation
translation modeladaptation
ASR SMT
language modeladaptation
investigate• conditions• models• adaptation techniques
(time-aligned slides)decide
Languages and Training Data
• ASR
24
Repository Language Acoustic Model Training Hours
Language Model Running Words
videolectures.netEnglish 768 4657 million
Slovenian 27 75 millionpoliMedia Spanish 390 1609 million
Language Pairs and Training Data
• SMT
25
Repository Language Pair Bilingual Sentences
Monolingual Running Words
poliMedia Spanish to English 12.9 million 57 million
videolectures.net
English to Spanish 12.9 million 34 million
English to Slovenian 4.7 million 378 million
English to French 4.0 million 119 million
English to German 2.2 million 621 million
Slovenian to English 1.1 million 57 million
Mixing media fragments
Media fragments
Media mixing is the process by which self-contained parts of media (fragments) are identified and exposed via media repository interfaces, so that consumers can access and re-use only the parts they are interested in.
Media Mixer Hub
AV Content Provider (1-n)
AV Content Demander (1-m)
1) AV material
analysis and annotation
2) Fragment Definition
3) Rights and Cost
Assignment
6) Search. Browsing
7) Rights and Cost
Assessment
8) Download
4) Fragment Upload
5) Clearing (Sell)
9) Composition of new AV materials
10) Clearing (Buy)
annotated & linked Media Fragments
Mediamixer (http://mediamixer.eu/)
• Community set-up and networking for the reMIXing of online MEDIA fragments
• to set up and sustain a community of video producers, hosters, and redistributors who will be supported in the adoption of semantic multimedia technology in their systems and workflows to build a European market for media fragment re-purposing and re-selling.
• http://community.mediamixer.eu/
Matterhorn (Opencast)
Matterhorn – Opencast http://opencast.org/matterhorn/home
Matterhorn basic facts
• Opensource (http://opencast.org/matterhorn/download)• Started July 2009• Official Ver 1.3.1, March 2012, Ver 1.4 in testing• Starting 2Y funds by:
• The Andrew W. Mellon: 1M US• The William and Flora Hewlett foundations: 0.5 US
• Now supported by institutions itself
Adopting organisationsUniversity College Cork University of Applied Sciences Osnabrück University of Bergen University of California Berkeley University of California Davis University of Cape Town University of Helsinki University of Manchester University of Nebraska-Lincoln University of New Mexico University of Osnabrueck University of Saskatchewan University Of The Arts London University of Vigo Vienna University of Technology Visionaire Campus Universidad Carlos III de Madrid Universidad Distrital Francisco Jose de Caldas
Ben - Gurion University Boise State University Entwine ETH Zurich Ghent university Loughborough University North-West University Northwestern University OBIS/Oxford Brookes University, Oxford UK Polytechnical University of Valencia (UPV) Reformed Theological Seminary Rice University Rochester Institute of Technology RRZE Uni-Erlangen Tel Aviv University Teltek Video Research The Institute for Global Outreach Developments International UNINETT AS
New and emerging models
Video journal - status
• Video Journal of Machine Learning Abstracts• Volume 1, 2, 3 – now preparing Vol. 4• 146 video abstracts, 15k views• Building up review committes
• PlanetData NoE - Video Journal of Semantic Data Management Abstracts • Volume 1, now preparing Vol.2
Feedback from the community Have separate tracks Have different selection criteria (no need to check sound, etc). Allow submissions of more multimedia-like presentations
Live streaming via Ncast
UNESCO collaboration
• VideoLectures Winner of WSA in 2009• March 2013 WSIS+10 Global Champion• VideoLectures.Net was selected as the winner in the “e- Science & Technology” category• 2 directors attended the Gala ceremony• meeting with Mr Qian Tang, Assistant Director-General for Education, UNESCO and
discussed:• Boost-up education in UNESCO member states• Provide a hub for MA, MSc, PhD video content exchange• Bring the latest research and development in Africa and globally to institutions in Africa• Potential to improve and facilitate mobile learning• Creation of AI training/research in developing countrires
• Visit of UNESCO director-general Irina Bokova in Ljubljana 9.4.2013
Already taken actions towards the future• Future (not so distant)
• Responsive learning environments• Learning companions• New academia/research• Speaker modelling
Links and contacts
Davor [email protected]
http://www.translectures.eu/[email protected]
Alfons Juan-CiscarUniversitat Politècnica de València (DSIC)
http://www.xlike.org/ Marko GrobelnikJozef Stefan Institute [email protected]
http://www.k4all.org/
http://mediamixer.eu/ Lyndon NixonSTI - [email protected]
Cronicle.com