richard baraniuk openstax courseware. courseware vision phase 1 – reinvent the textbook $$$

Richard Baraniuk

OpenStax Courseware

courseware vision

phase 1 – reinvent the textbook

$$$

open digital content• open ed publishing platform

established in 1999• 25,000 learning objects• millions of users per month

• library of 25 free and open college textbooks

• addresses “access gap” for disadvantaged students

• 17 ecosystem partners• 1099 adoptions, saving 300,000

students over $30M

courseware vision

phase 2 – personalize the course

courseware vision

phase 2 – personalize the course

1. broader access to high-quality courseware

2. improve learning using modern science (machine learning, cognitive science)

3. validation in real classrooms + research

goals

digital assessment

• in use at 12 colleges(Rice, Georgia Tech, Duke, UT El Paso, …)

• built-in research infrastructure

• integrated cognitive science principles(collaborators at Duke, UT-Austin, WashU)

flexible platform for practice, assessment, and learning research

learning principles

retrieval practice– retrieving information from

memory is not a neutral event; rather it changes memory

spacing– distributing practice over time

produces better long-term retention than massing practice

feedback– closes the learning feedback loop– must be timely

two-step answer process engages students in retrieval practice

spacedconceptpractice

timely, informativefeedback

research verification

• experiment at Rice 2012

• findings: students using cognitive science principles in OST scored ½-1 GPA point better than those using standard practice homework

flexible platform for practice, assessment, and learning research

content

learninganalytics assess and track student

learning progress by analyzing their interactions with content

content

contentanalytics

determine relationshipsamong content elements

learning/content analytics

classical approach – knowledge engineering– domain experts pore over content, assessments,

data, tagging and building rules– fragile, expensive, not scalable, not transferable

modern approach – machine learning– learn directly from data– automatic– robust, inexpensive, scalable, transferable

standard practice

Johnny

Eve

Patty

Neelsh

Nora

Nicholas

Barbara

Agnes

Vivek

Bob

Fernando

Sarah

Hillary

Judy

stu

den

ts

problems

questions(w/ estimated inherent difficulty)

concepts

studentknowledge

profile

87

55

23

93

62

Patty

data

ML AlgsCog Scipersonalizednext task

analyticsto instructor

feedback and analyticsto student

curriculum(re)design

personalizedlearning pathways

cognitive science research

machine learning

cycles ofinnovation

crossing the courseware chasm

The Mainstream Market

Technology

Enthusiasts

Visionaries

Pragmatists

Conservatives

Skeptics

crossing the courseware chasm

The Mainstream Market

Technology

Enthusiasts

Visionaries

Pragmatists

Conservatives

Skeptics

long term impact

“There is not such a cradle of democracy upon the earth as the Free Public Library”

building the personalized courseware library

of the future

sparfa

students

pro

ble

ms

sparse factor analysis

• Goal: using only “grade book” data

white: correct responseblack: incorrect responsegrey: unobserved

infer:

1. the concepts underlying the questions (content analytics)

2. each student’s “knowledge” of each underlying concept (learning analytics)

from grades to concepts

students

pro

ble

ms

data– graded student responses

to unlabeled questions– large matrix with entries:

white: correct responseblack: incorrect responsegrey: unobserved

standard practice– instructor’s “grade book”

= sum/average over each column

goal– infer underlying concepts and

student understanding without question-level metadata

students

pro

ble

ms

data– graded student responses

to unlabeled questions– large matrix with entries:

white: correct responseblack: incorrect responsegrey: unobserved

goal– infer underlying concepts and

student understanding without question-level metadata

key observation– each question involves only

a small number of “concepts” (low rank)

from grades to concepts

students

pro

ble

ms

~ Ber

statistical model

converts to 0/1(probit or logisticcoin flip transformation)

estimate of each student’s ability to solve each problem(even unsolved problems)

red = strong ability

blue = weak ability

students

pro

ble

ms

+

SPARse Factor Analysis

~ Ber

students

pro

ble

ms

+students

concepts

SPARFA

each problem involves a combination of a small number of key “concepts”

each student’s knowledge of each “concept”

each problem’s intrinsic “difficulty”

~ Ber

students

pro

ble

ms

solving SPARFA

factor analyzing the grade book matrix is a severely ill-posed problem

significant recent progress in relaxation-based optimization for sparse/low-rank problems

– matrix based methods (SPARFA-M)– Bayesian methods (SPARFA-B)

similar to compressive sensing

standard practice

Johnny

Eve

Patty

Neelsh

Nora

Nicholas

Barbara

Agnes

Vivek

Bob

Fernando

Sarah

Hillary

JudyJanet

questions(w/ estimated inherent difficulty)

concepts

studentknowledge

profile

87

55

23

93

62

technology architecture

marketing and adoption• research partners will co-develop

– Salt Lake Community College, University of Georgia

• pilot partners will field test– The Ohio State University, Auburn University, University System

of Georgia-Online Courses, Central New Mexico College, South Florida State College, Maricopa CC District, Tarrant County CC

• scale-up — key elements– fit into existing faculty/student workflow– build an ecosystem of affiliate partners– execute advertising and marketing campaigns– employ viral new media approaches– employ direct marketing and customer relationship

management system

• proven success 2012-2014