Computable-oriented Characteristics Representation of Online Learners from the Perspective of Heterogeneous Network

The accuracy of learner characteristics analysis has long been an issue in existing learner models. In order to enhance the accuracy and to get the multi-dimensional personalized characteristics of learners, a learner characteristics analysis based on the perspective of heterogeneous network was proposed in this paper. Network embedding method was adopted to realize the computability of learner modeling, in which multi-dimensional characteristics of learners have been divided into explicit characteristic database and potential implicit characteristic database. A heterogeneous learning network composed of learner social network, content network and association network among attributes has been constructed, which integrated different multi-dimensional interaction information for learner modeling. The analysis of heterogeneous learning network examples reveals that the computable method facilitates the extraction of specific learner characteristics and provides new methods for learner modeling.