Online Learner Categories Recognition Based on MoE-LSTM Model

Abstract

Online learner behaviour patterns are comprehensive indicators reflecting the learning status and learning outcomes of online learners, which have a guiding role in the design, implementation and improvement of online education in the post-epidemic era, but the methods for their identification still need to be improved. To improve the accuracy of online learner identification, a model of online learner behaviour identification based on improved long and short-term memory networks with deep learning is established. Firstly, the online learner behaviour characteristics are extracted according to the behavioural science theory, secondly, the online learner type is identified based on the "hybrid expert system-long and short-term memory network" model, and then compared with other identification models, and finally, the results are outputted by the progressive gradient regression tree GBRT in the stacking integration framework and validated using The results were validated using a ten-fold crossover. The results show that the method is effective in portraying online learners, and its accuracy and robustness are improved compared to other algorithms.