Designing the Authoring Experience for an Adaptive Platform on Open edX

Facilitated by: Dale Johnson, Arizona State University

Workshop Goals

  • √ √ √ √ √ Improving learner experience ---- models for implementation
  • Architecture design model
  • Identify trends in personalization
  • √ √ Authoring experience, steps.model — how to do it easier?
  • Share current efforts
  • search engine integration
  • student feedback  for improving learning
  • beyond MVP
  •  Learning analytics

Teaching online feels like a black box. Questions we need to anywhere

  • Who needs help ?
  • What do they need help with?

Students arrive with different levels of mastery in a given subject. Why not give those who can demonstrate mastery on day 1 the final exam?

Those who are ahead a bored in the traditional lecture modal.

We are moving from a mass production model to a mass personalization model.

  • music, health, mapping
  • what about education?
    • All students at the same time with the same lesson
    • It's the greek amphitheater brought forward

Personalization: Need to deliver the right lesson to the right student at the right time.

How can we move to dynamic model?

LMS: fixed → variable
Presentation Group → individual
Content: Common → Personalized

Adaptive systems at ASU.
College algebra, 58% success rate prior to 2016, now 2016 on 79%+.
Massive jump in success rate. +20.5%.

Able now to measure the different in success rate among various groups.

Other classes:

Intro to Biology (fall 2015 using CogBooks)
Microeconomics (Fall 2017 Cengage Learning Objects)

Key questions

What is adapting to the student?

  • Lesson sequence
  • Content selection

What techniques guide the adaptation?

  • Algorithm (analytics) -reccommendations
  • Assessment - rapid remediation
  • Association - relationships between lessons (Math knowledge map has been researched/documented for years)
  • Agency - student chooses (reflection helps students learn)

Udo: The way material is delivered to the student can have an impact as well (delivery techniques)

Related Learning science

Learning benefits from pedagogical techniques used in adaptive systems:

  • Formative evaluation (d=.90)
  • Acceleration (d=.88)
  • Effective Feedback (d=.73)
  • Meta-cognition (d=.69)
  • Mastery based learning (d=.58)
  • Concept mapping (d=.57)
  • Interactive Content (d=.52)

Lessons should be a singular learning objective.

Would be great to have open edx be a destination for creating open curriculum (knowledge maps, etc)

Multiple ways of presenting the same information

  • Learner Specificity
  • A recommendation algorithm

Lessons as learning loops. Start with assessment > fail see content or success continue to next lesson  (repeat until success)

do: learning content > Maybe this is personalized. or has multiple versions
while: (failing assessment)

A/B testing content

How to speed up learning what content is better for teaching a particular idea

Other companies use scale/personalization without owning any of their content. (Spotify, Airbnb, etc)

  • Course Configuration
  • Concept Chart (Lesson Library)
  • Content Curation

Maybe courses is not the thing we should think about anymore. We should be thinking about lessons. 

"Ed molecule" should have Objective, Content, Assessment.

Exercise dashboard in khanacademy. Where is open edX's lesson map.