Intelligent Learning Environments CAILE Gautam Biswas Department of Electrical Engineering and Computer Science Institute for Software Integrated Systems Vanderbilt University USA gautambiswasvanderbiltedu ID: 201555
Download Presentation The PPT/PDF document "Choice Adaptive" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Choice AdaptiveIntelligent Learning Environments(CAILE)
Gautam BiswasDepartment of Electrical Engineering and Computer Science Institute for Software Integrated SystemsVanderbilt University, USAgautam.biswas@vanderbilt.edu
Supported by IES CASL, NSF REESE, and NSF HCC awards
Interactive Event: Applications of Virtual Agents, Student Modeling, and Knowledge Engineering in EducationICCE ‘09, Hong Kong
www.teachableagents.org
Slide2
Collaboration with Dan Schwartz and the AAA Lab at Stanford UniversityVanderbilt Team membersJim Segedy, Brian Sulcer, Rod Roscoe,
Hogy Jeong, Roger TaylorICCE 2009: IE-5Acknowledgements2Slide3
Choice and adaptivity in CBLEsBackground – current systemsWhy choice and adaptivity?
Choice Adaptive Intelligent Learning Environments (CAILE)Conceptual FrameworkSoftware DesignDiscussion and ConclusionsICCE 2009: IE-53
Outline of TalkSlide4
State of the Art of CBLEs
(in contrast to programmed instruction, sequenced curricula, drill and practice)Intelligent TutorsKoedinger, vanLehn, Mitrovic, et al. – sequence of instruction pre-arrangedcorrective feedback during problem solving based on student model self explanation
Students become more efficient learners
Open Learner EnvironmentsDimitrova, Bull, et al. – students can see and discuss student model created by system feedback more transparent – heightened awareness students better at choosing learning topicsSome Choice ; Adaptive to learning performance & choice of topics
Background
CBLEs
ICCE 2009: IE-5
4Slide5
Coaching systemsBurton, Brown, Lesgold, Lajoie, Crews, Biswas, et al – intervene sporadically only when student seems to be stuck
help student overcome suboptimal solutionsChoice in problem solving steps supports some reflection during problem solvinge Learning and Online Course Management systemsBrown, Graf and Kinshuk, et al. – students can study topics in order of their choice – system adjusts to students’ learning choicesChoice of topics; system adapts to choices
ICCE 2009: IE-55
Background …Slide6
Simulation, virtual worlds, game-based environments Barab
, Dede,, van Joolingen, de Jong, et al. – Students construct hypothesis Solution process by explorationLots of choice, adaptivity limitedCollaborative environments (human, virtual agents)Chan and Chou, White, Graesser,
Rickel, Johnson, et al., Students learn through more natural social interactions Modalities of interactions more extensiveSome choice
and adaptivityICCE 2009: IE-56
Background …Slide7
Hypermedia environmentsStevens and Thadani – IMMEX project – choice of resources – analyzed using machine learning techniques to produce student patterns of choice
Choice related to learner performance; no adaptivityAzevedo, et al. – think alouds to study developmental differences + instructional scaffoldingWinne, et al. – protocols + computer logs to study learner characteristicsBiswas, et al. – SRL feedback based on students’ activity patternsStudents more aware of SRL strategies
ICCE 2009: IE-57
Background …Seems like it is a good idea to combine choice and adaptivity Slide8
Get away from “one size fits all” approachesPreparation for future learning
Choice provides students opportunities for developing self-regulated learning skills Self-regulation is positively correlated with later academic achievement (Zimmerman, 2001; Duncan, et al., 2007)Proper choice can increase (intrinsic) motivation for learning (Ryan and Deci, 2000; Blanchard and Frasson, 2004)ICCE 2009: IE-5
8Why choice?
How do we provide the right choices?Give students sufficient autonomy or “feeling of control”Slide9
Novice (less-experienced) students often use suboptimal learning strategies (downside of choice)May lack prior knowledge to determine a set of steps that will aid their learning
Self-judgment abilities not well developedOften poor at forethought Misevaluate the effort required to learn successfullyMay not be motivated to learn deeplyICCE 2009: IE-59
Why adaptive?
Adaptivity with autonomyscaffolding, guidance, detection and feedbackSlide10
Role of choice
Conceptual Framework: Structure choices for learnerNatural, intrinsic motivation, transparentExample choices: Topics to learn, Resources, Social interactionsSupport learner control and autonomy (implications for intrinsic motivation and individual differences)
Alternate choices are transparent in environmentOpportunity to observe students – reveals their approach to learning and problem solving
e.g., short cuts – studying to answer a few test questions versus gaining a deep understanding of domain Software Design Framework: Computational architecture
Reconfigurable
Adapt to user
ICCE 2009: IE-5
10
How to design Choice
Adaptive
CBLEs?Slide11
Role of adaptivity
Unguided choices can result in confusion and frustration for novice learnersCognitive load; poor SRL skills; loss of motivationOpportunity to help students overcome suboptimal learning strategiesCBLEs can scaffold productive student choicesMonitor/model students activity patterns
student performance (traditional student models)student behaviors and choices (meta-level, strategies)Provide push towards more optimal activities
Promote explicit awareness of effective SRL skillsICCE 2009: IE-5
11
Choice Adaptive CBLEs
Adaptivity needed to support individualized instructionSlide12
Implementing a Choice Adaptive Environment (CAILE)
The environment is a “fun fair”Different learning topics in different parts of the fairIn each part, there are different learning boothsBoss area: students can go to boss area to test integrated knowledge across booths in region.For multiple regions there is a Big Boss booth Beating this boss permits leveling up – going to a more complex learning topic
Additional booths Playing ticketsRedeeming tickets for prizes
Help booths12ICCE 2009: IE-5Slide13
Fun Fair Map for EcoLand
Tickets
Prizes
Help
Big Boss
Waterland
Airland
Fireland
Water Boss
Air Boss
Fire Boss
Type A
B
C
A
B
D
F
B
C
13
ICCE 2009: IE-5Slide14
Learning individual topicsChoice of ResourcesAgent Interactions
AssessmentsBoss boothsIntegrating knowledgeApplying knowledge to problem solving situationsDistraction BoothsFun, but little or no learningICCE 2009: IE-514
BoothsSlide15
New Interface
15ICCE 2009: IE-5Slide16
Components of the New InterfaceThe Agent Space
Agents notify user when they would like to speakUsers can initiate dialogue with the agentAgent behaviors: defined in terms of an event structure and rulesThe Conversation SpaceAgents and Students converse using dialog trees.Agent says something, and the student chooses how to respond. A choice often entails an action to be taken16ICCE 2009: IE-5Slide17
CAILE Software Architecture
ICCE 2009: IE-517
Activity Library
Activity
Activity 2
Activity 1
Activity 3
…
Activity 2
Activity 1
Activity 5
Activity 8
Activity 7
Activity 4
Activity 1
An
Activity
is something that a student engages in
Simulation
Problem solving tasks
Listening to lectures
Activities are self-contained
Activities occur within booths
Learning Task
Multiple activities that occur together in booth
Activities can be linked using events and glue code written in XML
Agents
Monitor student behaviors and performance
Role-dependent conversation with students
React to events – especially suboptimal strategies
Behavior Library
Composed of rules
mini pattern-matchers that “listen” for patterns of student actions in the Environment Platform
Generate compound events and agent behaviors
Environment Platform
Coordinates all communication between the Learning Task and the AgentsSlide18
Role playing charactersMentor, Peer, Teachable AgentsBehavior defined by rules
Mini pattern-matchers that “listen” for patterns of student actions in the Environment PlatformRepresents both a meaningful observation of student actions and a meaningful responseICCE 2009: IE-518
AgentsSlide19
Currently, we have a reconfigurable version of Betty’s Brain we are ready to sharehttp://www.teachableagents.org
; http://build.teachableagents.orggautam.biswas@vanderbilt.edu ; james.segedy@vanderbilt.edu Continue to conduct research in the area of supporting the development of SRL strategiesAnalyzing activity patterns, build student behavior modelsLink models to performance
Develop online methods for detection and feedback (hidden Markov Models – HMMs)Building “fun fair” systemContinue building choice and adaptivity in system
Dimensions of choiceAdaptivity and feedback – combine performance, methods, and strategiesSystematic links to affect and motivationCreate tools to aid in configuring CAILEsCreate user-supported libraries of agents, behaviors, and servicesProvide support for teachers in choosing, configuring, and deploying CAILEs for specific learning units
Provide report and analysis tools for teachers to aid assessments and classroom teaching
Summary/ Future Directions
ICCE 2009: IE-5
19