Research And Application Of Learner Evaluation Model Based On MOOC Data

The development of emerging information technologies such as big data,cloud computing and artificial intelligence has promoted the deep integration of Internet and education,further improved the level of education informatization,and spawned a new intelligent education ecosystem.In the new education mode,learners can break through the limitation of time and space,more convenient and fast access to educational resources and learning knowledge.However,in the face of a large number of educational resources on the learning platform,how to reasonably evaluate the learners and educational resources,so as to provide a matching mechanism so that learners can quickly retrieve the needed educational resources,has become a serious problem in the construction of the current intelligent education ecosystem.Based on the cognitive structure of learners,the cognitive diagnosis model models the potential relationship between learners’ knowledge level,test characteristics and the final answer results.Combined with the model solving algorithm and according to the answer results of the test questions,the knowledge level of the learners and the characteristics of the test questions can be evaluated.Aiming at the problems of traditional cognitive diagnosis model,such as single prior information and imperfect mechanism,this paper proposes a new high-order cognitive diagnosis model,cognitive and response model,to evaluate learners.The main innovations are as follows:(1)A joint compensation mechanism for ability and effort characteristics has been established to model the cognitive stage of learners.Cognitive process refers to the process of learners’ mastery of course knowledge.This paper establishes a joint compensation mechanism for the ability characteristics and effort characteristics to jointly influence the knowledge level of learners,and based on this mechanism,models the cognitive stage of learners.In addition,it also discussed how to quantify learners’ effort characteristics based on learning activity data,and use the quantified effort characteristics parameters as the extended model prior information.(2)The importance level of knowledge points to answer exercises is introduced to improve the modeling of learners’ reaction stage.The reaction process refers to the process in which learners use their own knowledge to answer exercises.In addition to considering the characteristics of test questions involved in traditional cognitive diagnosis models such as test questions guessing parameters and error parameters,this paper also introduces the importance of knowledge points to the answering exercises,and constructs the knowledge weakness parameters to the learner’s response stage.mold.(3)Finally,a personalized teaching aid system was developed.The system constructs individual and group portraits based on the evaluation results of the cognitive response model on the relevant data of MOOC courses,and presents the relevant characteristics of learners,knowledge,and test questions in a visual form for the reference of learners and teachers;on the other hand,according to The predictive value of the answering result of the learner’s test questions,realizes the function of recommending test questions according to the learner’s self-defined difficulty level.The collected MOOC data is randomly selected for experiments on the training set data ratio in the range of 40%-90%,and the results show that: based on the Cognitive and Diagnosis model to predict the learner’s test scores,the accuracy rate remains at77.4%-84.1%,compared with the DINA,HO-DINA,Fuzzy-CDF and NeuralCD models,the optimal prediction accuracy rate is 75.0%-82.0%,there is a significant improvement,and the robustness is better.