Research On Improving The Knowledge Tracing Model To Improve The Predictive Effect Of Learners' Knowledge

In recent years,online learning systems have become more and more popular.There are more and more learners in the process of online teaching for educators.Tracking the knowledge status of each learner and providing personalized learning guidance has become a huge challenge.On the other hand,learners in the online learning system self-find knowledge through various redundant information,resulting in diversification of learning resources and learning paths,but not necessarily effective.In order to solve the above problems faced by educators and learners,the introduction of the knowledge tracing model to automatically track the learner learning process can not only feedback the learner's knowledge to the educators,but also let the educators know more about each learner;It is also possible to infer learner's knowledge weaknesses and recommend appropriate learning paths and learning resources to learners.This study analyzes the development of the knowledge tracing model and the existing problems through the analysis of the relevant literature in 2008-2018,and summarizes the problems of the model in the prediction accuracy and teaching from the two levels of learners and data.On this basis,the KDD Cup 2010 education competition data set is used as the carrier to implement the data modeling and specific application analysis of the improved knowledge tracing model.Firstly,the forgotten knowledge tracing model is constructed to compare and analyze the prediction accuracy of the original knowledge tracing model,and the forgetting knowledge tracing model is used to explore the difference between the prediction accuracy of all data and partial data of the learner;Secondly,construct a personalized knowledge tracing model based on the learner's initial probability and learning probability to track learner's learning differences;Then,according to different segmentation basis,a segmented personalized knowledge tracing model is constructed,and the iterations of learning parameters of different segment learners are adjusted to compare the effects of displaying the segmentation model;Finally,the application of knowledge tracing model in learning environment is analyzed in detail.The experimental results show that:(1)the integration of forgetting and data volume factors in the knowledge tracing model has a positive impact on the prediction accuracy of the model;(2)The individualized processing and segmentation control operations of the learner's learning parameters in the knowledge tracing model can improve the model prediction accuracy;(3)The knowledge tracing model can perform learner performance prediction and time estimation.Finally,the paper summarizes the problems still existing in this research,and points out that the research direction of further exploration has certain practical significance.