Students’ Ability Evaluation Based On Bayesian Method And Research On The Algorithm Of Test Question Recommendation

In recent years,the concept of competency-oriented teaching has become a consensus in the education sector.The traditional teaching process takes the mastery of knowledge points as the teaching goal,which is difficult to reflect the students’ ability level.Meanwhile,The development of smart education and the increasing demand for personalized learning.How to use data mining and artificial intelligence technology to evaluate students’ ability level,as well as personalized test recommendation has become a research hotspot.Based on the concept of competency-based teaching,this thesis proposes a structured competency model to describe the teaching objectives of the course.The model describes the basic teaching objectives in the form of "knowledge keywords +competency dimensions".Among them,knowledge keywords are extracted from the course content by teachers according to their experience.The ability dimension is designed based on Bloom’s educational goal classification theory.Then,the structural and graphical expression is formed by using the correlation between the basic teaching objective units.Secondly,according to the curriculum objectives and educational measurement theory,a Bayesian based student ability evaluation method is proposed.The method can be divided into stepless evaluation and graded evaluation.Stepless evaluation is a method that uses Rasch model,introduces Bayesian theory,and uses the parameter value estimated by NUTS algorithm as the result of capability evaluation.In order to solve the problem of low accuracy of model parameter estimation caused by too low sample size,K-means algorithm is used to grade the difficulty of test questions.The conditional probability table is constructed according to the capability grading strategy,and the probability of each capability level is calculated through Bayesian formula.The capability level with the largest probability value is taken as the evaluation result.Then,an algorithm of test question recommendation based on Naive Bayes classification is proposed.The algorithm obtains the probability values of the test questions under the two categories of "recommended" and "not recommended" through naive Bayes,and recommends the test questions with the highest probability in the "recommended" category.Finally,the intelligent assessment system based on the above ability model and evaluation method is designed and implemented based on the online teaching and learning needs of university teachers and students,and is applied in the actual teaching process.