Research On Key Technologies Of Personalized Recommendation Based On MOOC

The development of large-scale online education platform MOOC makes more and more people participate in a variety of course learning.More and more scientists are beginning to research related technologies.Internet education represented by MOOC has many new features.It has rich and abundant educational resources.Users are being offered a much wider choice.However,massive educational resources also bring many new challenges,and users often face new problems of "information overload".Therefore,the personalized recommendation of educational resources has been developed.This article starts with the personalized recommendation of educational resources and extends the concept of educational resources to all the data resources that are helpful to users' online learning.This article is devoted to recommending the traditional teaching resources as well as the questions and answers in the MOOC forum as an educational resource to users who are in need.In this paper,we deeply analyzes the problem of learning resources which is difficult to label due to the imperfect extraction of keywords in metadata description of learning resources,and proposes a keyword automatic extraction algorithm based on word graphs and sentence graphs.This article then researches the forum data and focuses on solving the problem of missing the best answer in the MOOC forum.This article achieves the goal of recommending the best answer to the questioner.The work of this paper can be summarized as follows:1.An automatic keyword extraction algorithm based on word graph and sentence graph is proposed,which makes it convenient to label learning resources of standard description.2.Based on the keywords extracted above,this paper presents a method of establishing vector model of learning resources.Not only this,but also put forward the feasible scheme of establishing user interest model accordingly.Then this article realizes the personalized matching of user interest and learning resources,and achieve the purpose of personalized recommendation of learning resources.3.Based on the MOOC Forum data,to achieve the goal of recommending the best answer to the questioner,this paper presents a series of features that are applicable to MOOC forum data.Then we train these features on random forest classification model.Moreover,we obtain the important features that determine whether an answer is best answer and their impact factors respectively.