Automatic Recognition of Learner Types in Exploratory Learning Environments
Saleema Amershi, Cristina Conati · 2010
Exploratory learning environments (ELEs) provide facilities for student-led exploration ofatargetdomainwiththepremisethatactivediscoveryofknowledgepromotesdeeper understandingsthanmorecontrolledinstruction[32].Throughtheuseofgraphsand animations,algorithmvisualization(AV)systemsaimtobetterdemonstratealgorithm dynamicsthantraditionallystaticmedia,andtherehasbeeninterestinusingthemwithin ELEs to promote interactive learning of algorithms [15,34]. Despite theories and intuitions behindAVsandELEs,reportsontheirpedagogicaleffectivenesshavebeenmixed[8,34]. Researchhassuggestedthatpedagogicaleffectivenessisinuencedbydistinguishing studentcharacteristicssuchasmetacognitiveabilities[8]andlearningstyles[15,34].For example,somestudentsoftenndsuchunstructuredenvironmentsdifculttonavigate effectively and so they may not learn well with them [20]. Such ndings highlight the needforELEsingeneral,andspecicallyforELEsthatuseinteractiveAVs,toprovide adaptivesupportforstudentswithdiverseabilitiesorlearningstyles.Thisisachallenginggoalbecauseofthedifcultyinobservingdistinctstudentbehaviorsinsuchhighly unstructuredenvironments.Thefeweffortsthathavebeenmadetowardthisgoalmostly CONTENTS 15.1 Introduction ........................................................................................................................ 213 15.2Related Work ...................................................................................................................... 215 15.3The AIspace CSP Applet Learning Environment ......................................................... 216 15.4 Off-Line Clustering ............................................................................................................ 217 15.4.1 Data Collection and Preprocessing ..................................................................... 218 15.4.2 Unsupervised Clustering...................................................................................... 219 15.4.3Cluster Analysis ..................................................................................................... 219 15.4.3.1Cluster Analysis for the CSP Applet (k =2) .......................................... 219 15.4.3.2 Cluster Analysis for the CSP Applet (k = 3) .......................................... 221 15.5 Online Recognition ............................................................................................................223 15.5.1 Model Evaluation (k = 2) ........................................................................................ 224 15.5.2Model Evaluation for the CSP Applet (k =3) ......................................................225 15.6Conclusions and Future Work .........................................................................................226 References .....................................................................................................................................227 relyonhand-constructingdetailedstudentmodelsthatcanmonitorstudentbehaviors, assessindividualneeds,andinformadaptivehelpfacilities[8,31].Thisisacomplexand time-consumingtaskthattypicallyrequiresthecollaborativeeffortsofdomain,application, and model experts.