A Study on Data-Driven Teaching Decision Optimization of Distance Education Platforms

Abstract

Distance education requires the teachers’ teaching decisions to be innovative, thus it’s very meaningful to optimize the distance education elements, upgrade the teaching activity quality, and realize sustainable development. Existing studies generally make selections on distance education schemes based on empirical knowledge, however, since the decision parameters are often of poor time-efficient and prone to human caused errors, the efficiency and accuracy of the output decisions can hardly meet requirements. Therefore, to overcome these shortcomings, this paper aims to study the optimization of teaching decisions based on the teaching data of distance education platforms. At first, a hybrid neural network integrating the Bi-directional Long and Short Term Memory (Bi-LSTM) model and the Convolution Neural Network (CNN) was introduced into the Teaching Decision-making Optimization (TDO) model to capture the features of bi-directional time series of teaching decisions and build feature space with stronger expression ability. Then, a multi-objective TDO model was built based on fuzzy logic reasoning, which was then used to solve the problem during teaching decision-making that it’s difficult for multiple decision element combinations in distance education to meet standards at the same time. At last, experiments verified the validity and superiority of the proposed model