The Research On Recognition Of Learner's Confusion And It's Application In Performance Prediction

Data-driven learner modeling refers to the modeling of the learner's knowledge,behavior and emotions based on the learner's learning interaction data.It is a hot topic in the field of intelligent education and adaptive learning.The learner's learning emotion model is an important part of the learner model and affects the learner's learning effect.This study focuses on learning confusion,the most common learning emotion recognition problem,using machine learning technology,and proposes facial confusion-based learning confusion recognition methods;based on empirical research,explores the relationship between learning confusion and academic achievement.In the first study,the online evaluation experiment was designed.Twenty subjects were selected to collect photos of facial expressions when the learners were online.Through the preprocessing of photos,an automatic detection model of learners' confusion based on machine learning was established.The study found that the random forest has the highest detection accuracy of learning confusion,more than 70%,and has a good prediction effect.The naive Bayesian algorithm has the lowest detection accuracy for learning confusion.In the second study,the online assessment of learners is collected.Behavioral and perplexed emotional data,through statistical analysis of the relationship between learning confusion and test accuracy,found that in the people who are confused,the correct rate of test questions is low.At the same time,a machine learning prediction model is established to model learning confusion and academic achievement,and it is found that logistic regression is the best classifier among all models.Through the analysis and summarization of learning confusion test and performance prediction experiment,this study draws the following conclusions:(1)Random forest is the best learning confusion detection model in this study;(2)Learning confusion has strong correlation with academic achievement Among the people who are sexually and confused,the accuracy of the learners' correct answers is low.(3)By establishing a regression model for learners who are confused with learning,they can effectively predict academic performance.This research enriches the method of learning confusion detection,and explores the influence of learning confusion on academic performance.It can be used as a reference for learning emotions to integrate into intelligent teaching system and help to achieve personalized learning.