Transcript presentation slides

Designing CIspace: Pedagogy and Usability
in a Learning Environment for AI
S. Amershi, N. Arksey, G. Carenini, C. Conati, A.
Mackworth, H. Maclaren, D. Poole
What is
?
• A set of interactive algorithm visualization tools for
demonstrating the dynamics of common Artificial
Intelligence (AI) algorithms.
• Currently includes 9 Java applets for AI topics such as
graph searching, constraint satisfaction, deduction,
planning, machine learning, robot control and belief and
decision networks.
What are interactive algorithm
visualizations?
• Type of software
visualization.
• Use of
– images
– animation
– interface elements to
interactively demonstrate
algorithm dynamics.
Background
• Since 1980’s, hundreds of visualization
systems and repositories have developed
• Despite availability, such tools have not
been widely adopted
• Limited by:
– Pedagogical concerns
– Usability deficiencies
Overview of design process
• Iterative design process:
– Identify pedagogical and usability goals
– Design and implement features to achieve goals
– Revise choices in light of evaluations
Overview of design process
• Iterative design process:
– Identify pedagogical and usability goals
– Design and implement features to achieve goals
– Revise choices in light of evaluations
Pedagogical Goals
• P1 - Increase student understanding of AI
algorithms and underlying
representations
• P2 - Support different types of learners
• P3 - Motivate and generate interest
• P4 - Promote active engagement
• P5 - Support various scenarios of learning
P2 - Support Different Types of Learners
• Provide support for students with varying
learning styles.
• Provide support for novices, and continue to
provide support as a student’s expertise
increases.
• Account for individual learning pace.
P4 - Promote Active Engagement
• Support active construction of knowledge
and new understandings.
P5 - Support Various Scenarios of Learning
• Examples:
– in-class demonstrations
– assignments
– individual exploration
Usability Goals
• U1 - Easy to learn
• U2 - Straightforward and efficient to use
• U3 - Easy to integrate into a course
U3 - Easy to Integrate into a Course
• Making visualizations easy to adapt to:
– individual teaching approaches
– course content
– other course resources
Overview of design process
• Iterative design process:
– Identify pedagogical and usability goals
– Design and implement features to achieve
goals
– Revise choices in light of evaluations
Coverage
• Coverage of nine different AI topics
• Facilitates course integration by:
– reducing time and effort needed to find visualizations
for each new topic
– enabling CIspace to be used as a resource throughout a
course
Modularity
• Originally modularized based on
Computational Intelligence, by David
Poole, Alan Mackworth, and Randy Goebel
• Each applet is self-contained so can be used
to support other popular AI textbooks
• Helps to ease course integration by:
– giving instructors flexibility in choosing supporting
textbooks and other course resources
– giving instructors the option to select only those applets
that apply to their intended course syllabi
Interactive Simulations
• Multi-scaled stepping
mechanisms for control of
the simulation
• Features for exploring
different aspects in detail
• Supports active
engagement
• Enables students to learn at
their own pace
Sample Problems
• Each tool equipped with sample problems
• Helpful for beginner students.
• For instructors, this means less time
searching for examples.
Creation of New Problems
• Including:
– inputting new data
– creating new knowledge bases
– constructing new graphs
• Supports active engagement
• Supports more advanced students
• Enables instructors to create their own
problems for students.
Consistency
• Including:
– common applet layout
– common menu content and
layout
– similar graphical entities
– modes for creating and
solving
– analogous methods for
executing algorithms
• Minimizes learning
time and facilitates use
Table Summary
P1
P2
P3
P4
P5
U1
U2
√
Coverage and Modularity
Visual Representations
√
Interactive Simulations
√
Control of Algorithm Pace
√
Comparison of Algorithms
√
√
√
√
√
√
√
√
√
√
Sample Problems
Creation of New Problems
U3
√
√
√
√
√
√
√
Consistency
√
√
Help
√
√
√
• Each goal is supported by at least two design features. We argue that
this level of redundancy provides an adequate foundation for a robust
and reliable set of tools.
Overview of design process
• Iterative design process:
– Identify pedagogical and usability goals
– Design and implement features to achieve goals
– Revise choices in light of evaluations
Evaluation
• Feedback from users
• Usability inspection:
• User studies
– The applet is at least as effective in increasing
understanding as the traditional method of studying
sample problems on paper.
– Students liked studying with the applet significantly (ttest, p<.007) more than studying with paper sample
problems.
Conclusions
• Results and feedback about CIspace have
been encouraging.
• Visualizations can be effective for both
educators and students when designed to
support pedagogical and usability goals.
• CIspace:
– www.cs.ubc.ca/labs/lci/Cispace
• Questions?
Thank You!
Future Work
• We continue to update our tools in light of
results from our evaluations.
• Customizable applets
– user customizable
– author customizable
• Quiz features
• Adaptive help