Research On Personalized Learning Path Planning Based On Online Learning Behavior

Learning paths consist of a series of learning activities or learning content that lead learners to a specific learning objective.Traditional Massive open online course platforms provides online learning services by offering students various pre-defined learning paths with a fixed sequence.By ignoring learners’ knowledge background and learning ability,fixed-order learning paths often cause a series of challenges.One of these challenges is the high dropout rate in MOOC.One-size-fits-all learning materials are flawed in that they do not take into account different learners’ abilities.It is possible for learners to waste too much time on knowledge they are not capable of mastering or have already mastered.The fixed learning paths are not adapted to heterogeneous online learners and are therefore not conducive to improving the learning efficiency of online learners.It has been suggested that personalized learning paths planning can be used as an alternative solution tothe above problems.Personalized learning paths help to provide specific learning sequences for different learners based on their ability,knowledge background,and learning preference.Researchers have proposed various learning path planning algorithms and approaches over the past few decades.Currently,researchers currently consider personality traits like learning styles as important parameters when constructing personalized learning paths.However,in personalized path planning research,students’ objective online learning behaviors(e.g.,video watching,exercising,and interacting behaviors)are often overlooked.Reseaches has shown that learners’ intrinsic behaviors initiate their learning activities.The extrinsic behavioral performance plays an important role in revealing their intrinsic learning characteristics.Therefore,this study focuses on the important role of online learners’ extrinsic behavior in personalized path planning.The personalized path planning based on learners’ behavior proposed in this study is divided into three steps.In the first place,the current expert-centered manual approach to assessing the difficulty of knowledge points is time-consuming and labour-intensive and cannot be applied to large-scale educational data analysis.Based on existing research,this study constructs a score model that measures the difficulty of specific knowledge points.The input of this knowledge point difficulty model is the average learning performance of all learners who have studied the knowledge point,and the output is the difficulty level of knowledge point mastery.Second,this study constructs a learner status model based on learners’ online learning behaviors(video watching behavior and practice behavior)in order to decide their learning status.The main purpose of learning status model is to determine and update students’ mastery of knowledge points in real time.Learners’ video watching behavior and normalized practice test scores for a specific knowledge point are used as input parameters to dynamically assess learners’ learning status.The outputs are the four learning states of the learners,i.e.,“Unlearned”,“Unmastered”,“Insufficiently mastered”,and “Mastered” state.Finally,the personalized learning paths are planned according the learners’ specific learning status,the difficulty of the knowledge and the prerequisite relationship among knowledge.When learners complete each chapter,the learning status is updated based on their online learning behavior and practice test results.The previous knowledge points should be reviewed if a student is assessed as being “Unlearned”,“Unmastered”,or “Insufficiently masterd”.The learner will not be assigned to the next chapter until the learner’s status is assess to be "Mastered".It is based on MOOCCube X,a large-scale online open course database supported by the Knowledge Engineering Group of Tsinghua University and Xuetang X,one of the most popular MOOC platforms in China,to verify the feasibility of this personalized path planning method.The results show that the personalized path planning method can effectively provide personalized learning paths for students with different learning states.In addition,to validate the effectiveness of the personalized path planning,this study uses offline evaluation to assess the effectiveness of the personalized path planning.Three metrics were proposed to assess the effectiveness of personalized learning path planning,namely effective behavior rate,learning completion rate,and learning efficiency.The evaluation results showed that the effective behavior rate,learning completion rate,and learning efficiency of the personalized pathway student group were better than those of the regular student group.This shows that personalized learning paths based on online learning behaviors can help improve online learning.The study expects to be of some reference significance and value in developing the personalized path planning function of online platforms and personalized learning practices.