Adaptive Online Teaching System Based On Data Analysis

With the growing popularity of Internet,online educational websites are massively accessed by learners,which contributes to large amount of valuable educational data.Adaptive teaching system,which comes from the combination of computer technologies and online education,helps teachers with teaching plans dynamically by mining such data.It is widely used by many educational websites.In this paper,we look into two modules of an adaptive online teaching system.Student grouping module is based on action sequences from students.Students within each group share similar behavioral patterns,which supports teachers to make individual teaching plans for each group.Subjective answer scoring module gives machine score for subjective answers in homework or tests from students.The scoring system learns how average peer-reviewed scores come out,basing on how words are used and how answers are related to their questions.It aims at providing reliable scoring system and reducing repeated workloads of teachers.We use evolutional grouping algorithm with the student grouping task,which can find grouping results that are temporally coherent.It performs better than static clustering algorithms.In the task of answer scoring,we use a long short-term memory network to predict score ranges.We achieve the maximal accuracy of 54% in a four-class classification setting,which is better than the probability of guessing(25%)but still can be improved.