As people have become accustomed to non-face-to-face education because of the COVID-19 pandemic, adaptive and personalized learning is being emphasized in the field of education. Learning paths suitable for each student may differ from those normally provided by teachers. To support coaching based on the concept of adaptive learning, the first step is to discover the relationships among the concepts in the curriculum provided in the form of a knowledge graph. In this study, feature reduction for the target knowledge-concept was first performed using Elastic Net and Random Forest algorithms, which are known to have the best performance in machine learning. Deep knowledge tracing (DKT) in the form of a dual-net, which is more efficient because of the already slimmer data, was then applied to increase the accuracy of feature selection. The new approach, termed the optimal knowledge component extracting (OKCE) model, was proven to be superior to a feature reduction approach using only Elastic Net and Random Forest using both open and commercial datasets. Finally, the OKCE model showed a meaningful knowledge-concept graph that could help teachers in adaptive and personalized learning. INDEX TERMSDeep learning based knowledge tracing (DKT), dual-net, elastic net, feature selection, knowledge component (KC), least absolute shrinkage and selection operator (LASSO), random forest (RF).

Due to the global impact of COVID-19, the use of non-face-to-face learning is increasing. For non-face-to-face learning, it is important to create a learning path based on efficiency. This study introduces the hidden Markov model (HMM) as a method of creating a learning path known as a network in which knowledge concepts are arranged in order, that is, the path of experience that students may encounter in class. and it aims to improve the accuracy of path prediction by using a variable selection technique that includes least absolute shrinkage and selection operator (LASSO), and random forest (RF) before performing HMM. In addition, this study aims to show that the learning path based on higher-order concepts made of precedence relationships from HMM is more accurate than other candidate paths. As a result of using data shared by AI-hub (https://aihub.or.kr/), the performance of HMM when selecting relational pairs using LASSO, and RF was improved significantly, and the case of using HMM when evaluating the learning path consisting of higher concepts was excellent in terms of model goodness of fit compared to other models.