【作者】 张玥；

【导师】 黄宏涛；

【作者基本信息】 河南师范大学 ， 教育技术学， 2021， 硕士

【摘要】 随着互联网与教育的深度融合,教育个性化发展成为教育新趋势。在个性化教育要求下,教师要以学生为中心关注学生知识状态以便更好地引导学生学习,认知诊断为分析学生的知识结构提供了必要的技术手段。以AHM为代表的认知诊断模型需提前梳理好知识点间的关联关系,借助知识点关联关系对学生知识状态进行精确评估,常用于数学、物理、化学等知识点关联较密切的学科。但其应用于像英语、政治等一些知识关联关系松散型学科进行认知诊断时,诊断结果准确度无法保证。通过对大量国内外相关文献的研究,笔者发现SOM神经网络能够在不依赖于知识点关联关系的前提下对答题模式进行自组织聚类,实现对知识状态的分类与评估,以完成在知识点关联较松散学科中学生知识状态的认知诊断。因此,本文将SOM神经网络引入认知诊断,提出了基于SOM神经网络的教学认知诊断模型。其基本思想为:学生参与认知诊断测验,规范化学生答题结果后,将学生的答题模式输入至SOM神经网络,进行自组织竞争训练聚类,生成学生答题模式类别。在每个类别中,按知识点对类别中的测试项目进行切片处理,计算每个切片的正确率,切片的正确率即是对应知识点的正确率,将类别内各知识点正确率与评定标准正确率进行比较,生成类别属性模式,通过类别属性模式确定类别中学生的知识结构,实现对学生知识状态的认知诊断。基于SOM神经网络的教学认知诊断模型可以通过诊断类别知识结构来评估学生的知识状态和存在的认知困难,为教师开展补救教学提供支持。本研究将《大学英语视听说》课程作为教学内容,把基于SOM神经网络的教学认知诊断模型应用于实际教学以检验其有效性和可行性。通过收集前测、后测、问卷调查以及各方面的数据进行统计分析。分析得出:基于SOM神经网络的教学认知诊断模型可以有效提高在知识点关联较薄弱学科的认知诊断精准度,为教师开展针对性的补救教学,改善教学效果,以及帮助学生更好的个性化补救学习提供有效的支持;同时从诊断准确率、学习效果、满意度三个维度对学生进行问卷调查,均得到了学生较高的评价。更多还原

【Abstract】 With the deep integration of the Internet and education,the development of personalized education has become a new trend in education.Under the requirements of personalized education,teachers should focus on students’ knowledge status in order to better guide students’ learning.Cognitive diagnosis provides necessary technical means for analyzing students’ knowledge structure.The cognitive diagnosis model represented by AHM needs to sort out the relationship between knowledge points in advance,and accurately evaluate the student’s knowledge state with the help of the relationship between knowledge points.It is often used in subjects with closely related knowledge points such as mathematics,physics,and chemistry.However,when it is applied to cognitive diagnosis in subjects with loose knowledge relations such as English and politics,the accuracy of the diagnosis results cannot be guaranteed.Through the research of a large number of domestic and foreign related documents,the author found that the SOM neural network can self-organize and cluster the answer patterns without relying on the relationship between the knowledge points,realize the classification and evaluation of the knowledge state,and complete the knowledge points.Cognitive diagnosis of knowledge status of middle school students in loosely related subjects.Therefore,this paper introduces SOM neural network into cognitive diagnosis,and proposes a model of teaching cognitive diagnosis based on SOM neural network.The basic idea is: students participate in the cognitive diagnostic test,and after standardizing the results of students’ answering questions,the students’ answering patterns are input to the SOM neural network,and self-organizing competition training clustering is performed to generate student answering pattern categories.In each category,the test items in the category are sliced according to the knowledge points,and the correct rate of each slice is calculated.The correct rate of the slice is the correct rate of the corresponding knowledge point.The correct rate of each knowledge point in the category is evaluated with The standard correct rate is compared,the category attribute model is generated,and the knowledge structure of the students in the category is determined through the category attribute model,so as to realize the cognitive diagnosis of the student’s knowledge state.The teaching cognitive diagnosis model based on SOM neural network can evaluate students’ knowledge status and existing cognitive difficulties by diagnosing the category knowledge structure,and provide support for teachers to carry out remedial teaching.In this research,the course of "College English Audiovisual Listening and Speaking" is used as the teaching content,and the teaching cognitive diagnosis model based on SOM neural network is applied to actual teaching to test its effectiveness and feasibility.Statistical analysis is carried out by collecting pre-test,post-test,questionnaire survey and various aspects of data.The analysis shows that: the cognitive diagnosis model of teaching based on SOM neural network can effectively improve the accuracy of cognitive diagnosis of subjects with weaker knowledge points,carry out targeted remedial teaching for teachers,improve teaching effects,and help students better Personalized remedial learning provides effective support;at the same time,a questionnaire survey was conducted on students from the three dimensions of diagnosis accuracy,learning effect,and satisfaction,and all of them were highly evaluated by students.更多还原