Dedication
“Toourfamilies”
Chapter1:ArtificialIntelligence(AI)-enabledremotelearningandteaching usingPedagogicalConversationalAgentsandLearningAnalytics
AmaraAtif,MeenaJha,DeborahRichardsandAyseA.Bilgin
1.2.1Challengesanddisappointmentsofonlinelearningorremoteteaching
1.2.2ArtificialIntelligence,PedagogicalConversationalAgents, andLearningAnalyticsinsupportforremoteteaching-learning
1.3OurexperiencewithPedagogicalConversationalAgents
1.3.1BuildingVIRTAinUnity3D.......................................................................9 1.3.2VIRTAarchitectureanddeploymentenvironment
1.4.1AccesstoVIRTAbyteachingweek
1.4.2HoursofaccesstoVIRTA
1.4.3AccesstoVIRTAbyweekandhour
1.4.4AccesstoVIRTAbydayofweek
1.4.5StudentinteractionswithVIRTA
1.5Discussion............................................................................................................21
1.5.1UtilizingPedagogicalConversationalAgentswith
1.5.2IntegratingLearningAnalyticsandPedagogical ConversationalAgents
1.6Futuretrendsandconclusion
Chapter2:Integratingaconversationalpedagogicalagentinto theinstructionalactivitiesofaMassiveOpenOnlineCourse
RocaelHerna´ndezRizzardini,He´ctorR.Amado-Salvatierra andMiguelMoralesChan
2.1Introduction
2.3.1Researchgoalsandhypotheses
2.4Resultsanddiscussion
2.5Conclusionsandfuturework
Chapter3:ImprovingMOOCsexperienceusingLearningAnalytics andIntelligentConversationalAgent
TariqAbdullahandAsmaaSakr
3.1Introduction
3.2OnlinelearningandMOOCs
3.4LAICAintegrationinMOOCs:framework
3.4.1Proposedframework
3.4.2LAICA:systemflow
3.4.3LAICA:analyticsanddissemination
3.5LAICAintegrationinMOOCs:exampleimplementation
3.5.1Knowledgebasecreation
3.5.2Databaseintegration
3.5.3Testplan
3.5.4Learninganalyticsintegration
3.6LAICAintegrationinMOOC:impactanalysis
3.7LAICAintegrationinMOOC:findingsanddiscussion
Chapter4:Sequentialengagement-basedonlinelearninganalytics
XiangyuSongandJianxinLi
4.1Introduction
4.2.1Learninganalytics
4.2.2Predictingstudent’sperformance
4.2.3Students’learningengagement
4.3.1Notationandproblemstatement................................................................77
4.3.2SPENstructureoverview
4.3.3Engagementdetector
4.3.4Sequentialpredictor
4.3.5Lossfunction
4.4Experimentsandevaluation
4.4.1Experimentsenvironment..........................................................................81
4.4.2Experimentsettings
4.4.3Comparedmethodsandmetrics
4.4.4Evaluation
4.4.5Experimentresultsanddiscussion
Chapter5:Anintelligentsystemtosupportplanninginteractivelearning
5.1Introduction
5.2Theoreticalbackgroundsofintelligentsystemsinactivelearningsystems
5.3.1LearningobjectsinMoodlesystem
5.3.2Relationshiplearningobject:learningtopic
5.3.3Structureofobserverasactivitymodule
5.3.4Dataanalyticstechniques
5.3.5Generalizedsequentialpatternsoflearningobject ....................................97
5.3.6FelderandSilvermanlearningstylemodel .............................................102
5.4Resultsanddiscussion .......................................................................................102
5.5Conclusion .........................................................................................................107
Chapter6:Aliteraturereviewonartificialintelligenceandethics inonlinelearning
JoanCasas-RomaandJordiConesa
6.1Introductionandmotivations .............................................................................111
6.2Artificialintelligenceinonlinelearning.............................................................113
6.2.1Chatbotsandeducationalbots .................................................................114
6.3Ethicsinonlinelearning ....................................................................................115
6.4Ethicsinartificialintelligence ...........................................................................116
6.4.1Datascienceandethics ...........................................................................118
6.4.2Openingtheblackboxwithexplainableartificialintelligence
6.5Limitationsofethicsbydesignandhowmoralsystemscan overcomethem ..................................................................................................121
6.6Reflectionsandguidelinesforanethicaluseofartificialintelligence inonlinelearning ...............................................................................................123
6.6.1Datascienceandfairness ........................................................................124
6.6.2Transparencyandhonesty .......................................................................124
6.6.3Purposeofintelligentsystem ..................................................................125
6.6.4Technologicalexclusionanddigitaldivide
6.7Concludingremarksandfuturework
Chapter7:Transferlearningtechniquesforcross-domainanalysisof postsinmassiveeducationalforums
NicolaCapuano
7.1Introduction .......................................................................................................133
7.2Relatedworks ....................................................................................................135
7.3Textcategorizationmodel ..................................................................................138
7.3.1Word-leveldocumentrepresentation .......................................................139
7.3.2Overalldocumentrepresentationandclassification .................................139
7.4Transferlearningstrategy
7.5Experimentsandevaluation
7.5.1Textcategorizationperformance
7.5.2Transferlearningperformance
7.6Conclusionsandfurtherwork
Chapter8:Assistededucation:Usingpredictivemodeltoavoid schooldropoutine-learningsystems
FelipeNeves,FernandaCampos,VictorStro¨ele,Ma´rioDantas, Jose´ MariaN.DavidandReginaBraga 8.1Introduction
8.2.1Recommendersystems
8.2.2Predictivelearningmodels
8.2.3Learningstyles
8.3.1Comparativeanalysis
8.4PRIOREnsembleArchitecture
8.4.1Predictionprocess
8.5DPE-PRIOR:DropoutpredictiveEnsembleModel
8.5.1DPE-PRIORinaction
RinaAzoulay,EstherDavid,MireilleAvigalandDoritHutzler
9.1Introduction
9.3.1Reinforcementlearningmethods
9.3.2Bayesianinference-basedmethods
9.3.3Summaryofthecomparisonofadaptivemethods
9.4Methodologyandresearchapproach
9.4.1Theevaluationfunctionsusedinourstudy
9.4.2Adescriptionofthesimulatedenvironment
9.5Simulationresults ..............................................................................................193
9.5.1Topperformingalgorithms .....................................................................194
9.5.2Comparisonofalgorithmsovertime .......................................................195
9.5.3Comparisonofalgorithmsusingdifferentevaluationfunctions...............195
9.5.4Algorithmcomparisonfordynamicallychangedlevelofstudents ..........197
9.6Discussion..........................................................................................................198
9.7Conclusions .......................................................................................................200
Chapter10:Actor’sknowledgemassiveidentificationinthelearning managementsystem ........................................................................205
YassineBenjellounTouimi,AbdelladimHadioui,NourredineELFaddouli andSamirBennani
10.1Introduction .....................................................................................................205
10.2DiagnosisofhighereducationinMoroccoandcontributionofe-learning .......207
10.2.1Positionofe-learninginstructuringinformationand communicationtechnologyprojectsinMorocco ...............................208
10.2.2Theinteractiontraces ........................................................................208
10.2.3Learner’straces .................................................................................208
10.2.4Tutor’straces ....................................................................................209
10.2.5Administrator’straces .......................................................................210
10.2.6Researcher’straces ............................................................................210
10.2.7Learningmanagementsystemtracesfiles .........................................211
10.2.8Tracemanagementbybigdata..........................................................211
10.2.9Collectandloadingtracesprocesswithbigdatatools ......................213
10.2.10Thedatastorage ................................................................................215
10.2.11Dataanalysis .....................................................................................216
10.3Machinelearningevaluationtools....................................................................218
10.3.1Seekinginformation ............................................................................219
10.3.2Datavisualizationtools .......................................................................220
10.4E-learningmassivedata ...................................................................................220
10.4.1Theanalysisprocessofmassivetracesinamassiveonline opencoursediscussionforum...............................................................221
10.4.2Extractionknowledgebybigdataanalysisofmassivelogfiles ..........222
10.4.3Extractionknowledgebystatisticaldescriptiveanalysisoflogfiles ...223
10.4.4Semanticanalysisofbigdatainthediscussionforum ........................229
10.5Conclusion .......................................................................................................233 References .................................................................................................................233
Chapter11:Assessingstudents’socialandemotionalcompetencies throughgraphanalysisofemotional-enrichedsociograms...................239
EleniFotopoulou,AnastasiosZafeiropoulos,IsaacMuroGuiu, MichalisFeidakis,ThanasisDaradoumisandSymeonPapavassiliou
11.1Introduction
11.2State-of-the-artanalysis
11.2.1Sociometricassessmentapproaches
graphattributes ...................................................................................248
11.3.1Datacollection,processing,andemotional-enriched sociogramscomposition
11.3.2Socialnetworkanalysis
11.3.3ProvisionofinteractiverecommendationsforSocial andEmotionalLearningactivities
Chapter12:Anintelligentdistancelearningframework:
JohnYoon
12.3.1Modulestructure:ThreeLayers—BeginwithBasic, DiveintoAdvanced,andApplytoApplications .................................280
12.3.2Layerstructure:FourPanes—Slides,Videos,Summary, andQuizletPanes ...............................................................................281
12.3.3Answeringquestionspostedbystudentsonlayers: AnswerCollectionandRanking
12.4Learningassessmentandfeedback
12.4.1Assessmentquestiongeneration
12.4.2Smartphoneassessment .......................................................................287
12.4.3Question-drivenreorganizationofonlinecoursematerials
12.5Analytics:miningfromlearninggraphandassessmentgraph
12.5.1Assessmentdataanalyticsfordistanceteaching
12.6Simulationresult
Chapter13:Personalizingalternativesfordiverselearnergroups:
13.1Introduction
13.2Personalizationinonlineeducation:trends,systems,andapproaches
13.2.1Personalizationandaccessibility
13.2.2Readabilityasastartingpointforpersonalization
13.3Implementingreadabilitytools:generalstepsandsuggestions
13.3.1Precoursepreparation:contentanddisplayoptions
13.3.2Precoursepreparation:contentandtoolselection
13.3.3Facilitatingtooluse—encouragingtaskcrafting
13.3.4Evaluationofpersonalizationandtooluse
13.3.5Evaluationofperformance ..................................................................311
13.3.6Additionalimplementationconsiderations
13.4.1Practicalimplicationsforeducators
13.4.2Futureresearch
Chapter14:Humancomputationforlearningandteachingor collaborativetrackingoflearners’misconceptions
NielsHellerandFranc¸oisBry
14.1.1Resultsfrompreliminarystudies
14.1.3Chapterstructure .................................................................................327
14.2.1Onlinehomeworksystems ..................................................................328
14.2.3Text-andnaturallanguageprocessingin Technology-EnhancedLearning
14.2.4WikisoftwareinTechnology-EnhancedLearning...............................330 14.3Resultsfrompreliminarystudies
14.3.1Validityofsystematicerroridentification
14.4.1Thesemanticwiki
14.4.2Thepersonalassistant
14.4.3Thetextprocessingcomponent
14.4.4Theanalyticscomponent
Listofcontributors
TariqAbdullah CollageofEngineeringandTechnology,UniversityofDerby,Derby,United Kingdom
He ´ ctorR.Amado-Salvatierra UniversidadGalileo,GESDepartment,Guatemala,Guatemala
AmaraAtif UniversityofTechnologySydney,Ultimo,NSW,Australia
MireilleAvigal TheOpenUniversityofIsrael,Raanana,Israel
RinaAzoulay DepartmentofComputerScience,JerusalemCollegeofTechnology,Jerusalem, Israel
MammedBagher BusinessSchool,EdinburghNapierUniversity,Edinburgh,UnitedKingdom
SamirBennani DepartmentofComputerScience,MohammadiaSchoolofEngineering, MohammedVUniversityinRabat,Rabat,Morocco
AyseA.Bilgin MacquarieUniversity,NorthRyde,NSW,Australia
ReginaBraga ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearchGroup, FederalUniversityofJuizdeFora,JuizdeFora,Brazil
Franc¸oisBry LudwigMaximilianUniversityofMunich,Munchen,Germany
SantiCaballe ´ OpenUniversityofCatalonia,Barcelona,Spain
FernandaCampos ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearch Group,FederalUniversityofJuizdeFora,JuizdeFora,Brazil
NicolaCapuano SchoolofEngineering,UniversityofBasilicata,Potenza,Italy
JoanCasas-Roma FacultyofComputerSciences,MultimediaandTelecommunication, UniversitatObertadeCatalunya(UOC)—Barcelona,Barcelona,Spain
JordiConesa FacultyofComputerSciences,MultimediaandTelecommunication,Universitat ObertadeCatalunya(UOC)—Barcelona,Barcelona,Spain
Ma ´ rioDantas ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearchGroup, FederalUniversityofJuizdeFora,JuizdeFora,Brazil
ThanasisDaradoumis CulturalTechnologyandCommunication,UniversityoftheAegean, Lesvos,Greece
EstherDavid DepartmentofComputerScience,AshkelonAcademicCollege,Ashkelon,Israel
Jose ´ MariaN.David ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearch Group,FederalUniversityofJuizdeFora,JuizdeFora,Brazil
StavrosDemetriadis AristotleUniversityofThessaloniki,Thessaloniki,Greece
NourredineELFaddouli DepartmentofComputerScience,MohammadiaSchoolofEngineering, MohammedVUniversityinRabat,Rabat,Morocco
MichalisFeidakis DepartmentofElectricalandElectronicsEngineering,UniversityofWest Attica,Athens,Greece
EleniFotopoulou SchoolofElectricalandComputerEngineering,NationalTechnicalUniversity ofAthens,Athens,Greece
EduardoGo ´ mez-Sa ´ nchez ValladolidUniversity,Valladolid,Spain
IsaacMuroGuiu InstitutMartaEstrada,Barcelona,Spain
AbdelladimHadioui DepartmentofComputerScience,MohammadiaSchoolofEngineering, MohammedVUniversityinRabat,Rabat,Morocco
NielsHeller LudwigMaximilianUniversityofMunich,Munchen,Germany
RocaelHerna ´ ndezRizzardini UniversidadGalileo,GESDepartment,Guatemala,Guatemala
MatthewHodges TelefonicaEducationDigital,Madrid,Spain
DoritHutzler TheOpenUniversityofIsrael,Raanana,Israel
DeboraJeske SchoolofAppliedPsychology,UniversityCollegeCork,Cork,RepublicofIreland
MeenaJha CentralQueenslandUniversity,Sydney,NSW,Australia
AnastasiosKarakostas CERTH(CentreforResearchandTechnologyHellas),Thessaloniki, Greece
AllisonKolling SaarlandUniversity,Saarbrucken,Germany
KristijanKuk UniversityofCriminalInvestigationandPoliceStudies,Belgrade,Serbia
JianxinLi SchoolofInformationTechnology,DeakinUniversity,Geelong,VIC,Australia
EdisMekic ´ StateUniversityofNoviPazar,NoviPazar,Serbia
KonstantinosMichos ValladolidUniversity,Valladolid,Spain
MiguelMoralesChan UniversidadGalileo,GESDepartment,Guatemala,Guatemala
FelipeNeves ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearchGroup, FederalUniversityofJuizdeFora,JuizdeFora,Brazil
NadiaPantidi ComputationalMediaInnovationCentre,VictoriaUniversityofWellington, Wellington,NewZealand
GeorgePalaigeorgiou LearnWorlds,Limassol,Cyprus
PantelisM.Papadopoulos AarhusUniversity,Aarhus,Denmark
SymeonPapavassiliou SchoolofElectricalandComputerEngineering,NationalTechnical UniversityofAthens,Athens,Greece
TijanaPaunovic ´ SchoolofEconomics,Doboj,BosniaandHerzegovina
GeorgiosPsathas AristotleUniversityofThessaloniki,Thessaloniki,Greece DeborahRichards MacquarieUniversity,NorthRyde,NSW,Australia
AsmaaSakr CollageofEngineeringandTechnology,UniversityofDerby,Derby,United Kingdom
XiangyuSong SchoolofInformationTechnology,DeakinUniversity,Geelong,VIC,Australia
VictorStro ¨ ele ComputerSciencePostgraduateProgram,KnowledgeEngineeringResearchGroup, FederalUniversityofJuizdeFora,JuizdeFora,Brazil
StergiosTegos AristotleUniversityofThessaloniki,Thessaloniki,Greece
YassineBenjellounTouimi DepartmentofComputerScience,MohammadiaSchoolof Engineering,MohammedVUniversityinRabat,Rabat,Morocco
ThrasyvoulosTsiatsos AristotleUniversityofThessaloniki,Thessaloniki,Greece
CostasTsibanis GreekUniversitiesNetwork,Athens,Greece
IgorVukovic ´ UniversityofCriminalInvestigationandPoliceStudies,Belgrade,Serbia
ArminWeinberger SaarlandUniversity,Saarbru ¨ cken,Germany
ChristianWintherBech AarhusUniversity,Aarhus,Denmark
JohnYoon DepartmentofMathematicsandComputerSciences,MercyCollege,DobbsFerry, NY,UnitedStates
AnastasiosZafeiropoulos SchoolofElectricalandComputerEngineering,NationalTechnical UniversityofAthens,Athens,Greece
Foreword
Newtechnologiesandespeciallyinformationandcommunicationtechnologieshave penetratedalmosteverydimensionofsociety,includingeducation.Thepromisefor increasedefficiencyandalmostunlimitedeffectivenesshascharacterizedeachnewwaveof educationaltechnologies.DuringthecurrentCovid-19pandemiceducationaltechnologies havebeenthecornerstoneofremoteeducationandanessentialelementofeverydaylife. Andofcourse,thepotentialsynergybetweenartificialintelligence(AI)andeducationhas beenconsideredforseveraldecades.Expertandknowledge-basedsystems,accompaniedby powerfulmachinelearningtechniques,havepromisedadaptiveandpersonalizedlearning, orevenseamlesslearningacrossformal,nonformal,andinformalcontexts.Lastly,data sciences(DS)haveemergedasanotherrelevantcompanion,sinceinteractionsbetween humansandsystemscanbeeasilyregisteredandexploited,andthereforedatacanbe analyzedandeventuallysupportalleducationalstakeholders.Designing,deploying,and evaluatingappropriatealgorithms,tools,andsystemsusingallthesecomponentsisthe ultimategoalofmanyresearchersanddevelopers.
Suchatechnocentricapproachhasmostlydominatedthediscourseregardingtheprominent roleofinformationandcommunicationstechnologies,AI,andDSineducation.However, teachingandlearningcanbeconsideredashighlycomplexindividualandsocialprocesses embeddedinawidersocialcontext.Cognitiveandsocialprocessescannotbeeasily modeledanditisquitechallengingtofindtheadequateroleofsoftwareagentsinthewider educationalprocess.Designcannotbeeffectivelyaccomplishedwithoutastrong involvementorallrelevantstakeholders.Ontheotherhand,richdatamayberequired insteadofmassiveclickstreamsthatmaybeinterpretedthroughthelensofappropriate educationaltheories.Thedebateontheroleofeducationaltechnologieshasnotyetended, andeventuallywemayseeariseofnewrelevantandnecessarydiscoursesthatmayshed lightonthisevolvingfield.
Thisbookaddressesseveralissuesthathavebeenhighlightedabove.Conversationalagents mayinteractwithlearnersonaone-to-onebasisorsupportinggroup-levelprocesses.
Learninganalytics(LA)maybecalculatedandvisualizedindashboardsasameansfor learnerself-,co-,orsocially-sharedregulationorasasupportfordecision-makingby teachers,curriculumdesigners,oradministrators.Machinelearningtechniquescanbeused
inordertoderivepredictivemodelsregardingat-riskstudentsanddropoutrates.LA-based solutionsmayalleviatethescaleissuesfortheefficientsupportofteachingassistantsin massiveopenonlinecourses.Appropriateinstructionalapproachesmaybesupportedbyor complementAIapproaches.Severalpromisesandchallengesareintroducedandillustrated byrelevantresearchersinthisbook.
Consideringthecontributionsmadeinthisbookandaglobalviewofeducational technologiesorAIineducation,onecouldmentionsomeimportantissuesthatmightbe partofthecorrespondingresearchagenda:
•Significantconcernshavebeenexpressedregardingthetransparencyand trustworthinessofAI-basedmodelingandrecommendations,andevenmoreacutelyfor theteachingandlearningprocesses,wheresocialorcognitiveaspectsareinvolved.
•Similarly,theagencyofthehumans,andespeciallyteachersandstudents,isindanger andtheeffectivenessofthedesignedtoolsisquestioned.Thereforehumansmustbe broughtintheloop,orevenbeatthecenterofthedesignprocess,givingriseto human-centeredLAandAI.
•Severalcriticalvoiceshavebeenechoedintheliterature,andespeciallyduringthe Covid-19pandemic,regardingthetechnocentricviewofeducationaltechnologies, advocatingforanincreasedfocusonthesocialaimsofeducationandcallingfora renewedroleoftheteachers,asmediators.
•Sinceadvancesineducationaltechnologies,andespeciallyinLA,haveusuallybeen derivedinasinglefieldandmostlybytechnologists,aunifiedtheoreticalviewmight beconsidered,forexample,throughtheconsolidatedviewofLAthatbringstogether design,learningtheory,andDS.
•AI-basedsupporttolearnersandteachersshouldconsidernotonlythelearnerasan individual,butalsotakeintoaccountthecommunity-levelinteractionsandthegrouplevelteachingandlearningprocesses.Similarly,theinstitutionalanalyticsmightbe analyzedinconjunctionwiththeLAinordertoallowforawideradoptionandimpact.
•Well-knowntechniquesandapproachesinAImightbefurtherexplored,suchas processandtextmining,sentimentanalysis,ormultiagentsystems,alwaystakinginto accountthespecialanduniquecharacteristicsofteachingandlearning.
•Anintegratedviewshouldbesoughtregardingtheselectionanduseofhardandsoft sensors;theconstructionofanalytics,meaningfultoboththeoryandstakeholders;and theprovisionofmirroring,advising,andguidingactions.
•Professionaldevelopmentandcapacitybuildingactionsmightbestudiedandenacted, especiallyregardingteachersandlearningdesigners.Theseinitiativesshouldespecially takeintoaccountthebarriersforasmoothandpedagogicallysoundadoptionofthe tools,andconsiderthatultimatelytheeffectiveuseoftoolsisheavilyconditionedby thetechnologicalandpedagogicalknowledge,oreventheattitudesandbeliefsof teachers.
Ibelievethatrelevantrecentinterdisciplinaryresearchworkhasbeenpavingthewaytoa betterunderstandingoftheroleofAI,LA,andeducationaltechnologiesineducation,and eventuallyadvancingtowardabettereducation.Thisbookallowsforafurtherstepforward, althoughsignificantandchallengingworkisstillpendinginthiscomplexandhighly relevantfieldforsociety.
YannisDimitriadis
DepartmentofTelematicsEngineering,GSIC/EMICResearchGroup, UniversidaddeValladolid,Valladolid,Spain
Preface
Onlineeducationandespeciallymassiveopenonlinecourses(MOOCs)aroseasawayof transcendingformalhighereducationbyrealizingtechnology-enhancedformatsoflearning andinstructionandbygrantingaccesstoanaudiencewaybeyondstudentsenrolledinany onehighereducationinstitution(HEI).However,thepotentialforEuropeanHEIstoscale upandreachaninternationalaudienceofdiversebackgroundshasnotbeenrealizedyet. MOOCshavebeenreportedasanefficientandimportanteducationaltool,yettherearea numberofissuesandproblemsrelatedtotheireducationalimpact.Morespecifically,there isanimportantnumberofdropoutsduringacourse,littleparticipation,andlackof students’motivationandengagementoverall.Thismaybeduetoone-size-fits-all instructionalapproachesandverylimitedcommitmenttostudent studentand teacher studentcollaboration.
PreviousstudiescombineArtificialIntelligence(AI)-basedapproaches,suchastheuseof conversationalagents(CA),chatbots,anddataanalyticsinordertofacetheabove challenges.However,thesestudiesexploretheseandotherAIapproachesseparately,thus havinglessimpactinthelearningprocess.ThereforetheeffectiveintegrationofAInovel approachesineducationintermsofpedagogicalCAandlearninganalytics(LA)willcreate beneficialsynergiestorelevantlearningdimensions,resultinginstudents’greater participationandperformancewhileloweringdropoutratesandimprovingsatisfactionand retentionlevels.Inaddition,tutors,academiccoordinators,andmanagerswillbeprovided withtoolsthatwillfacilitatetheformativeandmonitoringprocesses.
Specifically,thebookaimstoprovidenovelAIandanalytics-basedmethodstoimprove onlineteachingandlearning,addressingkeyproblemssuchastheproblemofattritionin MOOCsandonlinelearningingeneral.Tothisend,thebookcontributestotheeducational sectoratdifferentlevels:
•Delivernewlearningandteachingmethodsforonlinelearning(withaspecificfocuson MOOCs),buildingonnoveltechnologiesincollaborativelearning,suchasCAandLA, thatarecapableofboostinglearnerinteractionandfacilitatelearners’self-regulation and-assessment.
•Demonstrateandvalidatethebuiltcapacityforinnovativeteachingandlearning methodsandmainstreamthemtotheexistingeducationandtrainingsystems,bythe
design,execution,andassessmentofpilotsthatorchestrateindividualandcollaborative learningactivities.
•Promotehighlyinnovativesolutionsandbeyondthestate-of-the-artmodelsforonline andMOOC-basedlearningandimplementationswiththeintegrationofAIservices, suchas,forexample,basedonCAandLA,tofacecurrentandfuturechallengesand forsustainableimpactononlineeducationalandtrainingsystems.
•Demonstrateandexemplifyefficientteachingtechniquesleveragingthepowerof analyzingdatageneratedbysmartAI-basedinterfaces,suchasthosepromoting interactionswithCAinlearningenvironments.
•DeepenourunderstandingofhowCAtoolscancontributetoincreasingthe transactionalqualityofpeers’dialogueand,consequently,thequalityoflearning,in varioussituations,suchaslearninginacademicsettingsandalsocorporatetrainingin businessenvironments.
Theultimateaimofthisbookistostimulateresearchfromboththeoreticalandpractical views,includingexperienceswithopensourcetools,whichwillallowothereducational institutionsandorganizationstoapply,evaluate,andreproducethebook’scontributions. Industryandacademicresearchers,professionals,andpractitionerscanleveragethe experiencesandideasfoundinthebook.
Thisbookconsistsof15chapters,startingwiththeIntroductoryChapterwheretheBook Editors,ledbyStavrosDemetriadis,presenttheEuropeanproject“colMOOC,”which supportstheeditionofthisbook.Theaimofthisleadingchapteristodescribetherationale oftheprojectmotivatedbytheissuesfoundinthecontextofMOOCs,whichprovidea powerfulmeansforinformalonlinelearningthatisalreadypopular,engaginggreat numbersofstudentsallovertheworld.However,studiesonMOOCsefficiencyfrequently reportonthehighdropoutratesofenrolledstudents,andthelackofproductivesocial interactiontopromotethequalityofMOOC-basedlearning.Theprojectproposesand developsanagent-basedtoolandmethodologyforintegratingflexibleandteacherconfigurableCAalongwithrelevantLAservicesinonlineeducationalplatforms,aimingto promotepeerlearninginteractions.TheauthorsclaimthatCAsappeartobeapromisingAI technologywiththepotentialofactingascatalystsofstudents’socialinteraction,afactor knowntobeneficiallyaffectlearningatmanylevels.Fromthisperspective,thechapter providesreflectionsonthefirstprojectoutcomesemergingfromfourdifferentpilot MOOCs.Earlyconclusionsanalyzethechallengesforintegratingateacher-configuredCAchatserviceinMOOCs,providehelpfulguidelinesforefficienttaskdesign,andhighlight promisingevidenceonthelearningimpactofparticipatinginagent-chatactivities.
Therestofthebookchaptersareorganizedintothreemajorareas:
PartI: IntelligentAgentSystemsandLearningDataAnalytics:Thechaptersinthisarea addresstheuseofpedagogicalCAsandLAtoprovidesupportive,personalized,
andinteractiveonlineteachingandlearninginlearningmanagementsystems (LMSs)andinparticularinmassiveeducationasinMOOCplatforms.Benefits andchallengesoftheproposededucationalstrategiessupportedbythese technologicalapproachesareunveiledandtheresearchresultsareillustratedwith practicaladoptionsinrealcontextsoflearning.Thecross-cuttingscopeofthe researchapproachescanbeappliedtodifferentknowledgeareasandlearning modesandstyleswiththeultimatepurposetoimproveandenhancetheonline teachingandlearningexperience.
PartII: ArtificialIntelligenceSystemsinOnlineEducation:Thisareastartswithexploring theintersectionbetweenAI,onlineeducation,andethicswiththeaimtodraw people’sattentiontotheethicalconcernssurroundingthiscrossroads.Therestof thechaptersintheareaprovideAI-basedsolutionstoaddressrelevantissuesfound incurrentonlineeducation,suchaspoorpersonalization,highacademicdropout, learners’disengagement,andlowparticipation,manyofthemresultingfrom facingonlineeducationatscaleandbigdata.Todealwiththeseissues,the chaptersproposetousedifferentAItechniques,suchasmachinelearning, sentimentanalysis,andnaturallanguageprocessing.Simulationresultsintermsof technicalperformanceandaccuracyarecomparedwithsimilarapproaches,and implicationsoftheseresultsforonlineeducationareillustratedintermsof improvingtheeffectivenessoftheonlineteachingandlearningprocessatscale.
PartIII: ApplicationsofIntelligentSystemsforOnlineEducation:Thechapterscoveringthis areapresenttheapplicabilityofdifferentapproachesofintelligentlearningsystems tovariousdomainsandforavarietyofpurposes,namelytheanalysisof socioemotionalprofileswithineducationalgroups,toovercometheuniformityof onlinelearningcontentstodealwithheterogeneouslearners,tosupportlearnerswith varyingreadingdifficulties,andtoimproveteachingandlearningofscience, technology,engineering,andmathematics(STEM)subjects.Strongimplicationsand furtherchallengesoftheapplicationoftheseapproachesincludemakingonline educationmoreeffective,multidisciplinary,andcollaborative,personalized,andfair.
Thechaptersinthefirstareaof IntelligentAgentSystemsandLearningDataAnalytics are organizedasfollows:
Atifetal.inChapter1,AI-EnabledRemoteLearningandTeachingUsingPedagogical ConversationalAgentsandLearningAnalytics,claimthattheadvancementsinAIhave potentiallycreatednewwaystoteachandlearn,suchastheuseofLAtomonitorand supportstudentsusingdatacapturedinLMSs.Tobackupthisclaim,theauthorsinthe chapterreportthebenefitsofusingAI-enabledCAsinmultipleunits/subjectsacrosstwo universitiesandillustratehowtheseCAscanplayarolesimilartoateacherorpeerlearner bysharingtheexpertisetheyhaveacquiredfromtheknowledgecontainedin student teachersocialinteractionsinLMSforumsandgrade-bookteacherfeedback.
Thechaptershowshowunliketeachersorpeers,theseCAscanbecontactedanonymously atanytime,theydonotmindbeingaskedthesamequestionrepeatedly,andtheycan empowerstudentstoexploreoptionsandoutcomes.Thechapterconcludeswitha discussionofthepotentialofLAtoautomateCAinteractions.
Chapter2,IntegratingaConversationalPedagogicalAgentintotheInstructionalActivities ofaMassiveOpenOnlineCourse,byRizzardinietal.addressesthetopicofusing pedagogicalCAstoofferawiderangeofpossibilitieswhenincorporatedintovirtual trainingcourses.ThechapterismotivatedinthecontextofMOOCs,wheretheinteraction withthestudentsisatscale,thushinderingpersonalizedinteractionbyhumanteachers.The authorsbelievethatanadequateconfigurationofpedagogicalCAshasthepotentialto providepersonalizedattention.However,theauthorsclaimthatthereareno“one-size-fitsall”approachesintermsofpedagogicalCAsgiventhattheconversationsusuallystartfrom scratch,withoutmuchusercontext,becomingespeciallyproblematicwhenaddressingthe issueofscalabilityinMOOCswherestudentsshowdifferentstatesandasimilarapproach isnotusefulforallofthem,requiringtostartwithapreviouscontext.Toaddressthisissue, theauthorsproposetheuseofLAtoprovideabettercontextfordecision-makingand initialvaluestolaunchthemodelresultingingreaterpossibilityofsuccess.Tothisend,the goalofthechapteristopresentaprototypeintegratingaCAembeddedintothe instructionalactivitiesofaMOOCwiththeultimateaimtoincreasemotivationandstudent engagementtoachievetheirlearninggoalswhileproducingimprovementsinstudents’ behaviorandhighercompletionrates.
AbdullahandSakrinChapter3,ImprovingMOOCsExperienceUsingLearningAnalytics andIntelligentConversationalAgent,discusstheeffectivenessthatonlinelearninghas provedinthelastyearsamongawiderangeoflearners.Inparticular,theauthorsclaimthat MOOCshaverevolutionizedtheshapeoflearningasasubstitutionaltoolcomparedtothe conventionaleducationalsystem,dueto,amongotherreasons,theirflexibilityintiming, eliminationofeconomicandgeographicalconstraints,whileenablinglearnersfrom differentculturestocommunicateandsharetheirknowledgethroughonlineforums.Then, theauthorsturnthediscussionintothechallengesfoundinMOOCsthatneedtobefaced, suchashigherdropoutsratesamonglearnersatdifferentphasesofthecourseandreduction inparticipationleveloflearners.Thechapteraimstoaddressthesechallengeswhile enhancingtheMOOCsexperiencethroughtheprovisionofaninnovativeframeworknamed LearningAnalyticsTechniqueandIntelligentConversationalAgent(LAICA)withthe purposeofintegratingLAandintelligentCAstoimprovetheMOOCexperiencefor learnersandeducators.ThechapterprovidesathroughoutdescriptionoftheLAICA frameworkfromthearchitecturalview,andacasestudyofimplementationandintegration oftheframeworkinaMOOCisprovided.
InChapter4,SequentialEngagement-basedOnlineLearningAnalyticsandPrediction, SongandLianalyzehowonlineeducationhasbecomeawidelyacceptedteachingmethod
overtherecentyearsasanintegratedlearningplatformprovidinglearningmaterialsand assessmenttools.Intheiranalysis,theauthorsclaimthatthroughthecompleteaccessrights andrecordsofthestudents’completeactivitiesonthelearningplatform,students’learning engagementandevaluationresultscanbewellanalyzedandpredicted.However,withthe developmentandchangesinteachingcontents,theauthorsclaimthatnewchallengeshave emergedasvariousnewformsoftextbooksandinteractivemethodshavebeenintroduced intovariousonlineeducationplatforms,whichmakemoreimplicitlearningpatternsbe learned,resultinginstudents’onlineactivitiesbeingcloselyrelatedtotheirfinalgrades. Fromthismotivation,theauthorssimulatelearningactivitiesinanewteachingformatin ordertoaccuratelypredicttheirfinalperformancebyleveragingimportantresearch outcomesintheLAfield.Eventually,thechapteraimstoexplainindetailhowtointegrate thelatestLAresearchmethodsintomodelingstudents’sequentiallearningengagement,so astoaccuratelypredictstudents’learningperformance.
Kuketal.concludesthefirstareaofthebookinChapter5,AnIntelligentSystemto SupportPlanningInteractiveLearningSegmentsinOnlineEducation,bydiscussingthe intelligenteducationalservicesusedbytoday’sLMSplatformsforthepurposeofcreating personalizedlearningenvironments.Theauthorsclaimthattheselearningenvironments mustbeadaptedforpersonalstudentleaningstyle.Tothisend,thepurposeofthechapter istopropose,develop,andexplaintheimp lementationofapers onalizedintelligent system.Theproposedsystemsuggestsadditionallearningresourcesthatwillsupport students’immersivelearningprocess,whichwillleadtowardbetteroutcomesoflearning activities.However,theauthorsconsiderthat onlinee-learningsystemsshouldimplement successfulmethodsandevaluationtechnique swhentakingdifferentteachingpathsthough facingtechnicalchallengess tillunsearched,whicharethemainmotivationofthechapter. Tothisend,theauthorsinthechapterurgethateverye-learningsystemshouldhave differentinteractivel earningsegmentsintheformoflearningobjectsintext,video, image,quiz,etc.,asentitiesineachsepar atecourseinthee-learningsystem,and supportedbyLAtechniquesasthemostappropriatemethodforautomaticdetectionof studentlearningmodels.Tothisend,thecha pterpresentsLAtechniquesforanalyzing learningpathscomposedfromfourdifferentle arningobjects,whicharethenimplemented inaMoodleenvironment.Asaresult,generalizedsequencepatternsaremapped,andan activitymodulenamedObserverisusedtotr ackstudents’learningbehavior.Theresults ofthetrackingareeventuallyusedtodevelopanintelligentsystemforplanning interactivelearningsegments.
Thechaptersinthesecondareaof ArtificialIntelligenceSystemsinOnlineEducation are organizedasfollows:
Chapter6,ALiteratureReviewonArtificialIntelligenceandEthicsinOnlineLearning,by Casas-RomaandConesadrawsattentiontohowAIisbeingusedinonlinelearningto improveteachingandlearning,withtheaimofprovidingamoreefficient,purposeful,
adaptive,ubiquitous,andfairlearningexperiences.However,theauthorsclaimthat,asit hasbeenseeninothercontexts,theintegrationofAIinonlinelearningcanhave unforeseenconsequenceswithdetrimentaleffectsthatcanresultinunfairand discriminatoryeducationaldecisions.Thereforethemainauthors’motivationisthatitis worththinkingaboutpotentialrisksthatlearningenvironmentsintegratingAIsystems mightpose.Tothisend,theauthorsexploretheintersectionsbetweenAI,onlineeducation, andethicsinordertounderstandtheethicalconcernssurroundingthiscrossroads.Asa result,thechapterprovidesanextensivereviewworkonthemainethicalchallengesin onlineeducationidentifiedintheliteraturewhiledistillingasetofguidelinestosupportthe ethicaldesignandintegrationofAIsystemsinonlinelearningenvironments.Theauthors concludethattheproposedguidelinesshouldhelptoensurethatonlineeducationishowis meanttobe:accessible,inclusive,fair,andbeneficialtosociety.
CapuanoinChapter7,TransferLearningTechniquesforCross-domainMOOCForum Postanalysis,addressestheroleofdiscussionforums,aspopulartoolsinthecontextof MOOCs,usedbystudentstoexpressfeelings,exchangeideas,andaskforhelp.Duetothe highnumberofstudentsenrolledandthesmallnumberofteachers,theauthorclaimsthat theautomaticanalysisofforumpostscanhelpinstructorstocapturethemostrelevant informationformoderatingandcarefullyplanningtheirinterventions.Tothisend,the authorfirstexploresseveralemergingapproachestotheautomaticcategorizationofMOOC forumpostsandclaimsthatsuchapproacheshaveacommondrawbackgiventhatwhen theyaretrainedonlabeledforumpostsfromonecourseordomain,theirapplicationon anothercourseordomainisoftenunsatisfactory.Forinstance,differentcourseshave differentfeaturespacesanddistributions,andcertainwordsmayappearfrequentlyinone course,butonlyrarelyinothers.Tohelpovercomethisdrawback,theauthorthen introducesacross-domaincorpus-basedtextcategorizationtoolthatincludestransfer learningcapabilitiesforthedetectionofintent,sentimentpolarity,levelofconfusion,and urgencyofMOOCforumposts.Theunderlyingmodel,basedonconvolutionaland recurrentneuralnetworks,istrainedonastandardlabeleddatasetandthenadaptedtoa targetcoursebytuningthemodelonasmallsetoflabeledsamples.Theproposedtool reportedinthechapteriseventuallyexperimentedwithandcomparedwithrelatedworks.
Chapter8,AssistedEducation:UsingPredictiveModeltoAvoidSchoolDropoutinELearningSystems,byNevesetal.discussestheimportantissueofstudentsdroppingoutof schoolasarealchallengeforeducationalspecialists,especiallyindistanceeducation classes,whichdealwithahugenumberofstudents’disengagementwithsocialand economiccosts.Inthiscontext,theauthorsclaimthatbehavioral,cognitive,and demographicfactorsmaybeassociatedwithearlyschooldropout.Motivatedbythisclaim, theaimofthechapteristoproposeanenhancedmachinelearningensemblepredictive architecturecapableofpredictingthedisengagementofstudentsalongwiththeclass.The systemnotifiesteachers,enablingthemtointerveneeffectivelyandmakestudents’success
possible,andstudentstogivethemachancetoturnback.Toevaluatetheproposed architecture,thechapterprovidesacasestudyshowingthefeasibilityofthesolutionand theuseofitstechnologies.Evaluationresultspointoutasignificantincreaseofgainin accuracyalongwiththeclass,reachingahighlevelofprecision.
Azoulayetal.inChapter9,AdaptiveTaskSelectioninAutomatedEducationalSoftware: AComparativeStudy,considerthechallengeofadaptingthedifficultylevelofthetasks suggestedtoastudentusinganeducationalsoftwaresystem.Intheirstudy,theauthors investigatetheeffectivenessofdifferentlearningalgorithmsforthechallengeofadapting thedifficultyofthetaskstoastudent’slevelandcomparedtheirefficiencybymeansof simulationwithvirtualstudents.Accordingtotheresults,theauthorsdemonstratethatthe methodsbasedonBayesianinferenceoutperformedmostoftheothermethods,whilein dynamicimprovementdomainstheitemresponsetheorymethodreachedthebestresults. Giventhefactthatcorrectlyadaptingthetaskstotheindividuallearners’abilitiescanhelp themincreasetheirimprovementandsatisfaction,thischaptercanassistthedesignersof intelligenttutoringsystemsinselectinganappropriateadaptationmethod,giventheneeds andgoalsoftheeducationalsystem,andgiventhecharacteristicsofthelearners.
Chapter10,Actor’sKnowledgeMassiveIdentificationintheLearningManagement System,byTouimietal.concludesthesecondareaofthebookbydiscussingonthe generationoftracesinanycomputersystemeitherbyuserinteractionswiththesystemor bythesystemitself.Theauthorsclaimthatwiththeproliferationofnewtechnologies, computertraceskeepincreasing,rapidlyandbrutallymakingenormouschangesinthefield ofeducationintermsoftechnicalmeansandteachingpedagogy.Inthecontextofonline education,theemergenceofMOOCsoffersunlimitedfreeaccessovertimeandspace whereinteractionsbylearnersgeneratelargeamountsofdatathataredifficultfortutors andlearnerstoprocessinlearningplatforms.Theauthorsfocusontheneedforlearnersto build,share,andseekknowledgeinaMOOCthroughdiscussionforums,whicharean efficienttoolforcommunication,sharingideas,opinions,andseekinganswerstolearners’ questions.Asacontributiontothisresearchfield,theaimofthechapteristoreportthe developmentofaframeworkcapableofmanagingbigdataindiscussionforumsinorderto extractandpresentrelevantknowledge,whichiscrucialinthecaseofMOOCs.The frameworkisbasedontheprocessofanalyzinglearners’tracelogfiles,whichincludesthe stagesofcollection,statisticalanalysis,andthensemanticanalysisoftracesoflearners’ interactions.Asstatisticalanalysisreducesthedimensionalityofthedataandbuildsnew variables,theauthorsproposetheLatentDirichletAllocationBayesianinferencemethodbe appliedtothreadsandmessagespostedinthediscussionforumsinordertoclassifythe relevantresponsemessages,presentasemanticresponsetothelearners,andenrichthe domainontologywithnewconceptsandnewrelationships.TheframeworkusestheApache Sparklibrariesforthecomputationspeedconstraints.