基于蚁群聚类算法的MOOC作业互评系统的分组研究

发布时间：2018-05-02 21:59

本文选题：MOOC + 蚁群聚类算法　；参考：《成都理工大学》2017年硕士论文

【摘要】：MOOC(massive open online courses),即大型开放式网络课程,MOOC平台一般都是面向全球或者某个特定范围群体,通过MOOC网络平台向学生传授知识正如火如荼、在全球范围内迅猛发展。具有优质的师资、优质的课程内容的结构体系是其主要特征。但MOOC平台也存在许多不可忽视的问题,比如学生资源的差异,学生的来源、学习工作背景、学习环境的差异等等,这些差异会直接导致学生在MOOC平台学习效果的差异。为促进MOOC环境下学生的学习质量,提升总体教学质量,实行学生间的作业互评是一项有效的措施。通过学生间的作业互评不仅可以有效的加深学生对课程内容的理解,对促进学生学习效果的共同提高具有良好的效果;并对解决MOOC环境下教师不可能对学生作业逐一批改的困境给出了一种有效的解决方法。学生间的作业互评如何分组?学生间如何进行互评?一种方式是随机的指定两名学生进行互评,但这种方式由于网络环境下学生彼此间的差异太大,可能导致出现不够理想的互评结果,学生互评成绩与实际成绩相差很大。本文提出一种基于学生背景大致相同情况下的作业互评模式,即进行互评学生的学习成绩、所在地域、学习环境等大致相同。解决问题的基本思路是将所有学生首先进行分组,也就是将背景大致相同的学生分成一组,然后在组内进行学生互评。这种作业互评分组模式虽然存在缺陷,但仍然不失为一种实际应用中可供参考的作业互评分组模式。在一定范围内,这种作业互评分配模式会使得学生所得成绩尽可能的接近学生的真实成绩。本文首先简要介绍了研究背景,和所运用方法的必要性。接着介绍了MOOC平台与蚁群聚类算法的相关基本理论知识及数学模型,第4章是本文重点,主要阐述了具体使用蚁群聚类算法来对学生进行分组;其中包括学生分组的问题,以及蚁群聚类算法的可行性,和具体分组实现等。第5章以某高校为例,给出了一个轻型MOOC平台的设计框架,以及学生作业互评信息系统的设计思路。最后给出了结论与建议。本文对MOOC平台下的学生作业分组研究具有实际应用价值;此外,在聚类过程中所给出的多值离散型属性的距离计算方法,使得离散属性量化后距离具有较高的均衡性,也是具有一定的实际应用意义。
[Abstract]:The MOOC(massive open online course platform is generally oriented to the global or a specific group. It is in full swing to impart knowledge to the students through the MOOC network platform, and it is developing rapidly in the global scope. With high-quality teachers, the structure of high-quality curriculum content is its main characteristics. However, there are many problems in MOOC platform, such as the difference of student resources, the source of students, the background of learning work, the difference of learning environment, and so on. These differences will directly lead to the difference of students' learning effect in MOOC platform. In order to promote the students' learning quality and improve the overall teaching quality under the environment of MOOC, it is an effective measure to carry out the homework evaluation among students. Through the students' homework evaluation, we can not only deepen the students' understanding of the course content, but also improve the students' learning effect. This paper also gives an effective solution to the problem that teachers can not correct students' homework one by one under MOOC environment. How to group the students' homework reviews? How to conduct mutual assessment among students? One way is to assign two students randomly to evaluate each other. However, due to the great differences between students in the network environment, there may be insufficient mutual evaluation results, and the difference between the students' mutual evaluation results and the actual results is very big. In this paper, a kind of homework evaluation model based on the same background is put forward, that is to say, the students' achievement, location and learning environment are roughly the same. The basic idea of solving the problem is to group all the students first, that is, to divide the students with roughly the same background into a group, and then to evaluate each other in the group. Although there are some defects in this model, it is still a reference mode in practical application. To a certain extent, this assignment model will make the students' scores as close as possible to the students' real scores. This paper first briefly introduces the research background, and the necessity of the methods used. Then introduced the MOOC platform and ant colony clustering algorithm related basic theoretical knowledge and mathematical model, the fourth chapter is the focus of this paper, mainly describes the specific use of ant colony clustering algorithm to group students, including the problem of student grouping, And the feasibility of ant colony clustering algorithm, and the implementation of specific groups. In chapter 5, taking a university as an example, the design framework of a lightweight MOOC platform and the design idea of student assignment evaluation information system are given. Finally, the conclusions and suggestions are given. This paper has practical application value to the study of student assignment grouping on MOOC platform, in addition, the distance calculation method of multi-valued discrete attributes given in the clustering process makes the distance after quantization of discrete attributes have a higher equalization. Also has certain practical application significance.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：G434;TP18

【参考文献】