结合先验信息的多属性诊断测验分类研究  

The Classification for High-Dimensional Cognitively Diagnostic Assessment Based on Prior Information

作　　者：叶子玉 秦春影 杨建芹 喻晓锋[1,3] 付道轩 YE Ziyu;QIN Chunying;YANG Jianqin;YU Xiaofeng;FU Daoxuan(School of Psychology,Jiangxi Normal University,Nanchang Jiangxi 330022,China;School of Mathematics and Information Science,Nanchang Normal University,Nanchang Jiangxi 330022,China;Key Laboratory of Psychology and Cognition Science of Jiangxi Province,Jiangxi Normal University,Nanchang Jiangxi 330022,China)

机构地区：[1]江西师范大学心理学院,江西南昌330022 [2]南昌师范学院数学与信息科学学院,江西南昌330032 [3]江西师范大学江西省心理与认知科学重点实验室,江西南昌330022

出　　处：《江西师范大学学报（自然科学版）》2023年第2期111-123,共13页Journal of Jiangxi Normal University(Natural Science Edition)

基　　金：教育部教育考试院“十四五”规划支撑专项课题(NEEA2021050);江西省社会科学基金(22JY16,21JY06);南昌市教育大数据智能技术重点实验室课题(2020-NCZDSY-012);江西省教育厅科技课题(GJJ2202013,GJJ2202018,GJJ212602,GJJ191691);江西省教育科学“十四五”规划课题(21YB257,21YB027)资助项目。

摘　　要：在动态的学习过程中,随着学习的深入和知识属性的个数逐渐增加,学生的知识状态也会发生动态变化.在这样的应用场景下如何结合先验信息提高诊断测验的判准率具有较大的挑战.该文提出基于学生对已学习属性的掌握概率来预测包含新属性后的属性向量的先验信息.考虑2种实际的应用情境,并通过模拟和实证研究来评价该方法的表现.结果表明:在2种实际学习情境中,结合先验信息在多属性诊断测验中能起到提高判准率的作用,其中在基于个体先验信息时的表现更好.基于预测先验信息的方法也摆脱了在以往研究中不同学习阶段测验属性个数一致或数量的限制,使得分类精度有较大改进.实证数据分析进一步表明该方法具有较高的应用价值.In a dynamic learning process,the student s knowledge state changes dynamically as learning progresses and the number of knowledge attributes increases.In such a scenario,it is challenging to combine a priori information to improve the accuracy of diagnostic assessments.The probability of student mastery of learned attributes is used to predict the prior information of the attribute vector after the inclusion of new attributes.Two realistic application contexts are considered and the performance of the method is evaluated through simulations and empirical studies.The results show that combining a priori information in two real-world learning contexts can be useful in improving accuracy in high-dimensional attribute diagnostic tests,with better performance when based on individual a priori information.The method based on predictive a priori information also breaks away from the limitations of previous studies in terms of the number of attributes that are consistent across different learning stages of the test,resulting in a greater improvement in classification accuracy.The empirical data analysis further demonstrates the applied value of the methods used in the methodology.

关 键 词：先验信息 诊断测验 多属性 纵向诊断测评 

分 类 号：B841[哲学宗教—基础心理学]