Invited lecture given at the University of Piraeus, focusing on a large scale case study of a learning technologies' application. Focused on the example of the Global Food Safety Partnership (GFSP, http://www.gfsp.org) and presented our view on backing it up with an infrastructure federating and linking different information sources/providers. These ideas have also been presented at this JALN paper: http://sloanconsortium.org/jaln/v17n2/open-and-scalable-learning-infrastructure-food-safety
11. Multi-Donor Trust Fund (MDTF) being
established to raise at least $45,000,000 for
implementation of a Roadmap and 5-year
workplan
aim: train small food producers, around the
world, using blended approaches
17. reflecting on complexity
#3
Program Facilitation
Open & Scalable
Open & Scalable
Learning Infrastructure
Learning Infrastructure
Includes : :
Includes
- -OER data pool
OER data pool
- -Course registry
Course registry
- -Curriculum Development
Curriculum Development
-Curricula Alignment
-Curricula Alignment
18. evolving the concept
• may aim higher than a single GFSP learning
platform – cannot generalize something that
requires a focused, regional approach
• rather develop a GFSP Learning Infrastructure
including (among others)
– Educational Offerings Aggregator Services
– Curriculum Support, Registry & Alignment Services
– GFSP Learning Portal (main front end)
– GFSP Learning Widgets/apps (to be integrated in web
sites, offered through smartphones/tablets, etc)
20. Educational Offerings Aggregator
• back end technology infrastructure
– ingest, harvest, aggregate course and OER
metadata from existing or new (e.g. legacy)
learning platforms and OER collections
– tools to allow course and OER providers to
align/map their metadata & classifications to
the GFSP ones
21. Curriculum Registry & Alignment
• representation of GFSP curriculum in
interoperable format (using learning outcomes,
competences)
– tools to allow other course providers to register and
express/map their curricula to the GFSP curriculum
– tools to facilitate the generation of multilingual
versions of the curricula descriptions
– generation of transformable curricula representations
to allow users to browse using preferred curicullum
format
22. GFSP Learning Portal
• main front-end to present project and allow
users to find information in the aggregated
sources
– various modalities (visual, device, thematic,
geographical, industry, …) for search &
discovery of courses and OER
– multilingual interfaces and metadata facilitated
by automatic translation engines
23. GFSP Learning Widgets/Apps
• search/discovery interfaces and mechanisms that
can be embedded in other web sites and portals
(widget-like or search pages in sites)
• mobile apps for various operational systems (iOS,
Android, Windows 8)
• back-end engine to allow straightforward
generation of adaptable versions of both
(thematic, industry, geographical, linguistic, …)
24. important distinction
• such a learning infrastructure is heavily
dependent on the back-end layers
• it is important to be able to power existing
applications and services
• the centralised portal mainly serves as
demonstrator
• will really change something if it provides a
wealth of resources around each topic
25.
26. a case study
• regional meat producer in Paraguay
– example scenario: exploring how their company
can start selling packaged cooked ham to an
international food distribution company
• product of high quality one, made from pure
pork ham
– let us assume that they would like to find out
more about the food safety standards of cooked
ham
27.
28.
29.
30.
31. this is why
#3
Program Facilitation
Open & Scalable
Open & Scalable
Learning Infrastructure
Learning Infrastructure
Includes : :
Includes
- -OER data pool
OER data pool
- -Course registry
Course registry
- -Curriculum Development
Curriculum Development
-Curricula Alignment
-Curricula Alignment
33. CONTENT PROVIDER
WITH UNORGANISED
COLLECTION
(e.g. Listed at Web
site or in DVD-ROM)
Chooses compliant tool
Metadata export in
Ingestion in
proprietary format & compliant tool
provides mapping
CONTENT PROVIDER
WITH CMS THAT DOES
NOT SUPPORT OAIPMH (e.g. Proprietary
DB)
CONTENT PROVIDER
WITH CMS THAT
SUPPORTS OAI-PMH
(e.g. FSKN compliant,
ePrints, DSPACE,...)
DOMAIN EXPERTS
publish & evolve
vocabularies &
ontologies
37. a. authoring/creation
• metadata creation is a painful and
costly process
– automatic generation can help
– high quality/accuracy/relevance
descriptions require human intervention
37
39. b. assurance/validation
• good online services demand high
quality (or at least not poor quality)
description of content
– someone needs to take the final decision
before something is published
– especially relevant when content
development has been costly/labourous
39
41. c. values/vocabularies
• mappings and crosswalks among
values and vocabularies of different
collections are crucial
– usually manually defined and maintained
– difficult to ensure that all applications
will publish and link their vocabularies
– vocabulary bank management tend to
become too complex for the purpose
that they serve
41
43. d. multilinguality
• for multilingual contexts, everything
needs to become (and be maintained)
multilingual
– metadata values and labels
– interface labels for various systems
• automatic translation helps but usually
produces rather rough/poor
translations
43
50. competencies description
• what I would like to learn is what you
need me to know…
– …but what is really needed is connecting a
job profile to the relevant course offerings!
54. expected learning outcomes
• what I am going to learn should be what I
am expected to know for my job…
– …but sometimes it’s not very clear what
this is going to be!
55. CerOrganic Curriculum description
DICLA training center
Capabilities: When completing this course you will
be able to perform basic routine operations in a
defined hydroponic context under close supervision.
56. more issues…
• old-fashioned legacy systems still used in such
traditional settings
– terms like “OER” and “MOOC” sound like science fiction
• novel technologies such as semantic stores and
ontology editing/managing environments are not
user-friendly and proven
– especially for such technology-ignorant users
• very rich semantics to be represented, handled and
exploited; but we are not there yet
58. targeted domain
• rich in data-oriented problems and
cases
• focused on “real” users
• inter-disciplinary work
• results related to societal
goals/challenges
59. increase use & reuse
• digital sources and collections
material to be used (and
potentially re-used) in several
contexts
– even different than originally
expected/thought of