The Research And Application Of Adaptive Testing In Tutoring System

Compared with the traditional teaching methods,online education provides learners with more free and convenient learning environment,as well as more abundant teaching resources.However,there are still many problems in the implementation of the online education platform in China.Most of the platforms are simply pushing learning resources to learners,and lack of consideration for the ability and knowledge background of different learners.Therefore,how to assess the ability and knowledge of learners in the process of learners’ learning and provide reasonable learning suggestions are of great practical significance for improving the quality of online education.Based on the basic idea of adaptive learning,this thesis evaluated the learner’s ability and knowledge by adaptive testing.In the process of adaptive testing,the system continuously evaluates the learner’s current abilities and selects the items most suitable for their current abilities for their tests.If the learner’s test result on the current knowledge point doesn’t meet the assessment requirements and the test termination conditions are met,the system will terminate the test and provide the learning interface of the corresponding knowledge points to learners.Otherwise,if the assessment requirements are met,the system will provide the test and learning interface of the follow-up knowledge points,ultimately to achieve the goal of guiding learners to targeted learning by the way of adaptive testing.First,this thesis studies the item response theory in computer adaptive testing,and adopts the three-parameter Logistic model in the item response theory as the main theoretical model.Next,this thesis analyzes and designs the tutoring system,and builds the corresponding domain knowledge model and learner model.Then,it focuses on the ability estimation method and item selection method in adaptive testing.By using the Monte Carlo method to simulate the adaptive test and comparing the performance of the two common ability estimation methods: maximum likelihood estimation method and Bayesian expectation posterior method,the maximum likelihood estimation method is chosen as the ability estimation method of this thesis,and improve its defects about may not to converge in some situation,Monte Carlo experiments show that the convergence of the improved capacity estimation method is obviously better than that before the improvement.In the aspect of item selection,an item selection method based on maximum information method is designed,this method takes into account the amount of the item information,exposure control,content balance and the correlation between knowledge points,it ensures the accuracy of the test while controlling the use of high-frequency items,thereby to improve the utilization rate of item bank.In addition,this thesis also designs a method based on classical test theory for setting initial values of item’s parameters,as well as a simple and effective maintenance method for item’s difficulty parameters to ensure the accuracy of item’s parameters during the test.Finally,this thesis implements the tutoring system based on adaptive testing,and integrates it with the online education platform to test and analyze the function of the adaptive test module and the question management module.The results show that the system implemented in this thesis meets the design requirements and achieves the desired design goals.