The MOOC Discussion Forum Discusses In-depth Evaluation Model Research

The depth of online discussion has a huge impact on the quality and effectiveness of learners' online learning.However,there is not much research on the depth of discussion in the MOOC discussion forum,especially the evaluation of the depth of discussions in the MOOC discussion forum.At present,the evaluation of online discussion mainly adopts content analysis method and scale method.These methods are to perform manual sampling analysis after the end of the course to predict the whole sample,and it is not possible to comprehensively,timely and automatically analyze the massive discussion posts in the MOOC discussion forum.This study uses the big data analysis method to construct a discussion depth evaluation model to effectively evaluate the online discussion depth of MOOC learners,so as to achieve accurate prediction and timely intervention.This research takes the MOOC platform of China University as the research object.Firstly,it assumes that the key factors affecting the depth of online discussion in MOOC forum are the type of questions,the type of knowledge,the response and duration of questions,and the role of curriculum participation.Secondly,analyze the depth of online discussion in MOOC forum.Current situation:The depth of online discussion in this course is obviously in the stage of shallow discussion and moderate discussion,and the communication between learners is not deep enough.Thirdly,the relationship between different types of influencing factors and different stages of discussion depth is verified by Chi-square test and Pearson correlation coefficient method.The statistical results show that the depth of discussion in MOOC discussion area is related to the type of' questions,the number of replies and the number of questions asked.The number of respondents,teachers'participation,duration of questions and types of knowledge points are positively correlated,and the correlation coefficient is large.These factors have a greater impact on the depth of online discussion.There is a weak correlation between the problem definition,the importance of knowledge points and the depth of online discussion.Both of them have little influence on the depth of online discussion in discussion area.The specific implementation of this study is as follows:(1)Through the pre-screening experiment of characteristic variables,six characteristic variables are selected:question type,knowledge point type,question duration and response umber,presiding over the questioner and teacher participation.(2)Establish the evaluation index system of MOOC discussion area discussion depth evaluation model.(3)Establishing the evaluation model:Applying the feature set to the four classification models of Logistic Regression,Decision Tree,Linear Support Vector Machine and Random Forest,adjusting the parameters of the evaluation models and selecting different feature variables to train the best classification effect of each evaluation model.The experimental results show that:(1)The feature set constructed in this paper is the optimal feature set of each evaluation model.The performance of the decision tree model and the random forest model on the optimal feature set is compared with logistic regression and linear support vector machine classification model.More superior,the accuracy,accuracy,recall and F value of the decision tree and the random forest model are all above 80%.(2)The ranking results of the importance of the six characteristic variables are basically the same in each evaluation model.Among them,the problem type,the problem duration,the number of problem responses,and the problem hosting situation are the four most important characteristic variables that affect the depth of discussion in the MOOC discussion area.