《最新：以诊断学习困难为考量之测验和评量系统文档资料》由会员分享，可在线阅读，更多相关《最新：以诊断学习困难为考量之测验和评量系统文档资料（30页珍藏版）》请在金锄头文库上搜索。
1、Relevant ResearchqGwo-Jen Hwang (2003), “A conceptual map for developing intelligent tutoring systems”, Computers & Education, Vol. 40, No.3, pp. 217-235. (SSCI) qGwo-Jen Hwang, Jia-Lin Hsiao and Judy C.R. Tseng (2003), “A Computer-Assisted Approach for Diagnosing Student Learning Problems in Scienc
2、e Courses”, Journal of Information Science and Engineering, Vol. 19, No.2, pp. 229-248. (SCI Expanded, EI)qGwo-Jen Hwang (2005), “A Data Mining Approach to Diagnosis Student Learning Problems in Science Courses”, Journal of Distance Education Technologies, Vol. 3, No. 4, pp.35-50. (EI) 1A conceptual
3、 map for developing intelligent tutoring systems (2003)qConventional testing systems simply give students a score, but dont give them the opportunity to learn how to improve their learning performance.qStudents would benefit more if the test results could be analyzed and hence advice could be provid
4、ed accordingly.qThis study proposes a conceptual map model, which provides learning suggestions by analyzing the subject materials and test results.2Concept Effect Relationships (CER)qMcAleese (1994, 1998) indicated that students learn new concepts and new relationships among previously learned conc
5、epts. qSalisbury (1998)that indicate the effect of learning one concept on the learning of other concepts.qex: The names and abbreviations of chemical elements and their atomic weights must be thoroughly learned to comprehend scientific writings or chemical formulae.qSuch conceptual relation has bee
6、n defined as “concept effect relationships” by Hwang (2003).3Conventional subject materials Structure qSubject materials was viewed as a tree diagram comprising chapters, sections, sub-sections and key concepts to be learned.4New Structure: Concept effect Relationshipsconceptual mappreviously learne
7、d conceptsHi-level conceptsNext, how to get the “Concept effect graphs” ?5Concept effect table (CET)qTwo-dimensional table.qIf CET(Ci,Cj)=1, it is said that Ci is one of the prerequisites of Cj.“Division”的的prerequisites concepts (NPj)=2Positive integersSubtractionMultiplication6QiCjC1C2C3C4C5C6C7C8C
8、9C10Q1Q2Q3Q4Q5Q6Q7Q8Q9Q10SUMERRORER(Cj)5000010000610.16=1/61400000000500=0/50230000000530.6=3/50015000000610.16=1/60020500000720.28=2/70000040002640.66=4/60000005000640.63=5/80020020001540.8=4/50000000140540.8=4/50000000050551.0=5/5ER(C1) ER(C2) ER(C3)Illustrative example of a test item relationship
9、 table (TIRT)test item relationship table (TIRT) wrong wrong wrong wrongTIRT(Qi, Cj): ranging from 05, 0:no relationshipsSUN(Cj): total strange of Cj (Cj)ERROR(Cj)SUM(Cj)-7Concept effect graphC1C9C8C7C6C2C4C5C3C10= The acceptable error rateER(Cj): Add in “To-Be-Enhanced learning path”: average error
10、 ratio of Cj for the students who get the bottom 50% of test scores.=0.558PATH 3: Addition Multiplication Division Prime numbersWeight=Max(ER(C3), ER(C7), ER(C9), ER(C10)=1.0PATH 1: Addition SubtractionNegative integersWeight=Max(ER(C3), ER(C6), ER(C8)=0.8PATH 2: Addition Subtraction Division Prime
11、numbersWeight=Max(ER(C3), ER(C6), ER(C9), ER(C10)=1.0C1C9C8C7C6C2C4C5C3C10最關鍵的補救路徑：最關鍵的補救路徑： PATH 2 & PATH 3 = Max Weight9Fuzzy output for learning guidanceqFuzzy sets on ER(Cj)ER(Cj)0 0.5 1.0 Well- LearnedVery well- LearnedMore or less well-Learnedpoorly-LearnedVery poorly- LearnedMore or less poor
12、ly-Learned1.00.500.30.550.2ER(C1)=0.16ER(C1)=0.16 Very well-learned0.3Well-learned0.55More or less well-learned0.210Illustrative example of a learning guidance (To Student)ConceptLearning status of the conceptC1Zero You have learned the concept well.C2Positive integers You have learned the concept v
13、ery well.C3AdditionIt seems that you more or less misunderstood this concept.C4Odd You have learned the concept well.C5EvenYou have learned the concept well.C6Subtraction.It seems that you misunderstood this conceptC7MultiplicationIt seems that you more or less misunderstood this concept.C8Negative
14、integersIt seems that you seriously misunderstood this concept.C9Division It seems that you seriously misunderstood this concept.C10Prime numbers It seems that you seriously misunderstood this concept.misunderstood concepts “Addition”, Subtraction, Negative integers, Division, Multiplication and Pri
15、me numbers,11Comments for the studentqComments for the student:1. According to the diagnosis from the system, we found that you have misunderstood concepts Subtraction, Negative integers, Division, Multiplication and Prime numbers, which perhaps results from the misunderstanding of Addition. In othe
16、r words, the major learning problem of yours is the misunderstanding of concept Addition, which affects the learning of other concepts.2.Suggestion: enhance the study in 1. Addition Subtraction Division Prime numbers and2. Addition Multiplication Division Prime numbers sequences.PATH 2PATH 312Practi
17、cal Example13Intelligent testing and diagnostic system (ITES)qWindows NT platform.qCLIPS format. (a well-known expert system shell developed by NASA (Giarratano & Riley, 1989)qITES comprises: lstudent profile databaselitem banklJava-based interfaceltesting and diagnostic unitlfuzzy interface.WWW and
18、 Java-Based User InterfaceFuzzy InterfaceExpert SystemInference EngineTeachersStudentsWWW BrowsersKnowledgeBaseItem BankStudentProfileTestingandDiagnosticUnitSystemLog14ExperimentqExperiment period: 2001.9.2001.12. (3 months)qMaterial: An elementary schools natural science course.qConditions: qThe s
19、ame teacherqSixty K-6 students from two classesqGroup-A (Control group): 30 students, qreceived regular on-line tutoringqtesting without learning guidanceqGroup-B (Experimental group) : 30 studentsqreceived regular on-line tutoringqlearning suggestions and relevant homework15Pre-testq樣本變異數同質性t檢定t-te
20、st =0.05, t(29)=1.699Group-AGroup-BGroup diff.(1-2)Grade Pooled Equal 58 2.32Grade Satterthwaite Unequal 56.7 2.32Variable Method Variances df t ValueEquality of variancesGrade Folded F 29 29 1.36 Group B: performanceof Groups A and B in the pre-test differs significantly. Variable Method Num df Den
21、 df F value Pr F t2.32 t (=.05)1.699the performance of Groups A and B in the pre-test differs significantly16Post-testt-test =0.05, t(29)=1.699Group-AGroup-BGroup diff.(1-2)GRADE Pooled Equal 58 -2.47GRADE Satterthwaite Unequal 56.7 -2.47Variable Method Variances df t ValueEquality of variancesGrade
22、 Folded F 29 29 1.95 0.0782 Variable Method Num df Den df F value Pr F Group B Group A: performanceof Groups A and B in the post-test differs significantly. t2.47 t (=.05)1.699a significant difference between the performance of Groups B and A17A graphical user interface is provided for constructing
23、the conceptual effect graph18A Computer-Assisted Approach for Diagnosing Student Learning Problems in Science Courses (2003, SCI Expanded, EI)qUnfamiliar with computer programming, time-consuming, friendly user interface for teachers to apply it unaided.qCER Generator: Generating concept effect rela
24、tionships based on previous records of student answer sheets19CER-builder- Define support/belief values to generate a set of CER 20The generated CER can be edited by the teacher21Physics courseqThe final concept effect graph with support = 0.2 and belief = 0.9.Fig. 8. Percentage of concepts involved
25、 in the concept effect relationships of the Physics course.22Evaluation of the Efficacy of the CER BuilderqExperimental period: March 2001 to June 2001.qMaterial: Natural Science course at an elementary schoolq60 studentsqGroup A (Control group)(V1): received regular on-line testing without learning
26、 guidance.qGroup B (experimental group)(V2): received learning suggestions and relevant homework after each on-line test.qPre-test, post-test: The statistical results obtained by applying SPSS.23Post-test & Pre-testp-value(Sig.2-tailed) = .024 (=.01)=Ho成立the mean score of Group A = Group B.p-value (
27、Sig.2-tailed) =.009 Ho不成立the mean score of Group B Group A24A Data Mining Approach to Diagnosis Student Learning Problems in Science Courses (2005, Journal of Distance Education Technologies)qA data mining approach that is capable of assisting teachers to provide information needed for guiding stude
28、nts during the learning process.qData Mining for Constructing Concept-Effect Relationships.qInput Data-Answer Sheet Summary Table2526Similar experimental results have been derived on three courses27Evaluation for efficacy of the data mining approach28ConclusionqThis study proposes a conceptual map m
29、ethod for modeling the prerequisite relationships among concepts to be learned.qSeveral experiments have been conducted, which indicate that the group of students who received the learning guidance can make significant progress compared with the control group.qTwo different approaches have been proposed to assist the teachers in constructing the concept-effect relationships. qMore experiments on Social Science or Language courses might be interesting29Thank you for your attention!30
