Research And Implement Of Uniform Grouping Students Problem In Massive Open Online Courses

With the development of the Internet,the emergence of large-scale online courses website and related mobile applications have broken the time and space limitation of traditional teaching,creating a good environment for the transmission of high quality teaching resources.Although the course videos can be repeatedly used in online teaching,some other teaching affairs might not,such as divideding students into groups for group learning or team project.These affairs all rely on the good understanding of the online course participators.However,the large amount of students and the lack of interaction between teachers and students seriously impede teachers from such online teaching affairs,which took precious time of teachers but did not achieve the ideal teaching effect.Aiming at this problem,this thesis studies the grouping students problem in large-scale online courses,proposed two novel algorithm:grouping algorithm based on the floating range and grouping algorithm based on the storing function,which reduce the teacher's online teaching task and promote the process of full automation online courses without teachers.Specifically,this thesis first quantifies the benefits after the grouping according to the theory of cooperative learning.Then formally define the student uniformity grouping problem and prove it a NPC problem.Lack of efficient accurate solutions,this thesis proposes an iterative grouping framework based on the genetic evolutionary algorithm to converted the student grouping uniformity problem to finding exchangeable students problems in each iteration.In order to improve the solution efficiency of iterative grouping framework,the score floating profit function and the average floating profit function of each iterative calculation process is analyzed.Based on the monotonicity and iterative computational characteristic of the average floating profit function,the grouping algorithm based on floating range is put forward which successfully reduce the time complexity of common iterative grouping algorithm from O(kn~2(n/m))to O(kn~2)by using binary search and recursive calculation skills which make it possible to get the score floatable range by only calculating a point.But its computing only rely on the qualitative analysis of profit improvement,which makes it still improvable.Therefore,the grouping algorithm based on the storing function was proposed.It analyze the iterative redundant computation in the process of grouping students in detail.Combined with the storage features of the average floating profit function,the grouping algorithm based on the storing function uses a small amount of storage space to eliminate a lot of redundant computation.In each round of iteration,it find the global optimal exchangeable student pairs analyzing the profit increment quantitatively,which reduce the time complexity of common iterative grouping algorithm from O(kn~2(n/m))to O(kn~2lg(n/m))and ensure the rapid and stable convergence of the algorithm.Finally,we evaluate these two grouping algorithm in both real and artificial datasets,the experimental results show the high efficiency of the proposed grouping algorithm based on floating range and grouping algorithm based on storing function,and the gains of the grouping results were superior to most other grouping algorithm.Moreover,this ascension of grouping profit has nothing to do with the specific distribution of the scores.Besides,this thesis also designed a grayscale matrix diagram for the visualization of the grouping results,which validates that the grouping results of proposed algorithms have benefits equilibrium among each small groups,and there is an obvious dividing line between the leaders and the followers.