This presentation outlines work done by Enovation and research partners in the Percolate project to investigate allowing for personalised JIT help in Moodle.
2. Context – Percolate Project
• e-Learning Industry working with academic
researchers
• Industry defined challenging use cases
• Challenge is to apply research to use case
• Third level use case – helping students with
problems
Industry Academia
3. The Problem
• Students working online needs help
• Finding wood from trees difficult on WWW
• I need help with X now!
– Understand subject
– Pedagogically sound
– Personalised to me
photo by London College of Fashion
4. The Problem
• NDLR (DSpace) Moodle search
• One size fits all
• Celebrate individuality
– Prior knowledge
– Type of resources
– Course context
5. Recommender system
• Hooks Moodle into a learning resource
repository
• Search understands what peers found useful
• Search understands the conceptual structure
of subject
• Uses conceptual structure to understand
learners needs
• Results organised according to a pedagogical
strategy
8. How does it work?
• Semantic search
– Based on conceptual structure of subject area
• Social search
– Based on whether “similar” learners found a
resource useful
• Composition engine
– Compose resources from search into learning
episode
9. User Trial
DCU Trial
• 18 DCU 4th year Mechanical Engineering students
• 6 week trial (complete)
• Students used application in the context of the
following assignment:
“You are a process engineer for a multi-international
institution that wishes to introduce an advanced
manufacturing technology for one of their new advanced
material based products…”
10. User Trial
DCU Trial
• 527 completed search sessions
• 419 results selected (130 distinct results)
• User feedback:
– 36 results rated or tagged
– 34 post-confidence scores for concepts
“A wider range of information needs to be
“Good system, still in its early uploaded in order for it to accomodate
stages” [Student A] the academic objectives of the materials
module” [Student B]
11. Next Steps
• Percolate LTC
• Better UI
• Manual effort
• Non-intrusive methods of gaining info
12. Questions?
Acknowledgements:
Dr. Dermot Brabazon – DCU
Catherine Bruen – NDLR
We kindly acknowledge Enterprise Ireland for their
support for this project.
Contact: mark.melia@enovation.ie
Contact: mark.melia@enovation.ie
Notas do Editor
UCD, Trinity College
Saw the talk yesterday – we have worked with the NDLR to allow for searching NDLR in MoodleWe wanted to see if the search could be specific to learner needs Move away from one size fits all -> celebrating differences in learners [prior knowledge, type of resources] deliver resources that are
Saw the talk yesterday – we have worked with the NDLR to allow for searching NDLR in MoodleWe wanted to see if the search could be specific to learner needs Move away from one size fits all -> celebrating differences in learners [prior knowledge, type of resources] deliver resources that are
Search looks for concept in conceptual structureSystem tries to establish if concept not known due to missing pre-requisite knowledgeReturns material based on learner intent, knowledge levels of concepts and associated metadata on learning resources Generates a learning episode– composition engine using a pedagogical strategy – intro, examples, summary