DynamicSolutionGraph(DSG)-SlideTutor动态的解决方案图（DSG）-slidetutor

《DynamicSolutionGraph(DSG)-SlideTutor动态的解决方案图（DSG）-slidetutor》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、ReasoninginaTutoringSystem:TransformingKnowledgetoTeaching.OlgaMedvedevaCenterforPathologyInformatics,UniversityofPittsburghJuly6,20047thInternationalProtégéConferenceOutlineOurapproachforteachingvisualdiagnosisGeneralsystemarchitectureKnowledgerepresentationindifferenttutormodulesPlusesandminuseso

2、foursystemJuly6,20047thInternationalProtégéConferenceMedicalKBTrainingSystemChallengeProblemsMedicalknowledgeiscomplexanddynamicErrorsinKBcancauseseriousproblemsDemandsonextendibilityandmaintenanceoflargeKBSRequirementsCombineknowledgerepresentationandflexibleinstructionalsystemAdaptivefornewobserv

3、ablesanduniquestrategiesReusableandmodularJuly6,20047thInternationalProtégéConferenceIntelligentTutoringSystemsParadigmITSstrivetoreplicateamethodofteachingandlearningexemplifiedbyaone-on-onehumantutoringinteractionModelTracingITSguideuserthroughproblemspace,cancorrecteachsmallintermediatereasonin

4、gstepCognitiveTutorsbasedonACT_Rtheoryoflearningproceduralizedeclarativeknowledgeintherules(stepinstructions)July6,20047thInternationalProtégéConferenceStudentModulePedagogicKnowledgeInterfaceExpertModelAllowcorrectstepsCorrecterrorsGivehintsonnextstepCollectdataonwhatstudentdoesMakepredictionsonwh

5、atstudentknowsProvidedataforpedagogicdecisionmakingCanvasforproblemsolvingMakegoalsvisibleCasesequenceWhentointerveneHowtointerveneIntelligentTutorSystemStructureJuly6,20047thInternationalProtégéConferenceDisadvantagesofITSParadigmDevelopedforhighlyproceduraldomainsNotdesignedforlargecomplexdynamic

6、declarativeknowledgeDomainspecificproductionrulesknowledgerepresentationMaintenanceisdifficultandtimeconsumingKnowledgemodificationaltertherulesJuly6,20047thInternationalProtégéConferenceSlideTutorCharacteristicsSlideTutor–asystemtoteachvisualclassificationproblemsolvinginPathologySimilartoothermed

7、icaldiagnostictasksCombinationofsearch,identification,interpretationWellcharacterizeddiagnosticreasoninginmedicaldomainsSomeareasarehighlyalgorithmic,some–notBothempiricalandtheoreticalworkcanguidethedevelo