ScholarWorks@중앙대학교

초록

This study proposes the analysis model using HMM that can estimate the learning state hidden from observations in a continuous sequence in order to reveal the prerequisite relations among the knowledge concepts(KC). Also, the problem domain was compressed with meaningful KC relations by applying ElasticNet(LASSO) and Random Forest(RF), which are regression coefficient reduction methods. As a result of applying elementary mathematics evaluation data to the ElasticNet(LASSO)+RF+HMM model, the average accuracy is improved by 7% compared to the previous study applying MSMM method. This study presents a model framework of the KC prerequisite relations that can make probabilistic correction for students' guesses or mistakes and contributes in that personalized education guides the appropriate learning path according to the learner's knowledge status by improving the knowledge concept relations.

키워드