Article: Real-time effectiveness evaluation of online education based on LSTM-transformer model Journal: International Journal of Reasoning-based Intelligent Systems (IJRIS) 2025 Vol.17 No.12 pp.24 - 34 Abstract: Online education, with its flexibility, has become an integral part of the education sector. To address the challenges posed by existing research, which struggles to capture spatio-temporal locality and handle lengthy historical evaluation sequences, this paper first inputs historical evaluation data into a long short-term memory network (LSTM) to discover long-term sequential relationships in the evaluation data. The LSTM's output is then fed into the Transformer encoder, followed by an encoding layer that feeds into the transformer layer, where multi-head attention mechanisms enhance concurrent learning of long-term dependencies. Second, the final evaluation prediction results are obtained through a softmax output. Finally, an improved Bayesian optimisation algorithm is used for hyperparameter iteration, and the optimal hyperparameters for the evaluation model are selected. Experimental outcome demonstrates that the average evaluation accuracy of the proposed model has improved by 5.98%-12.24%, validating the efficiency of the proposed model. Inderscience Publishers - linking academia, business and industry through research

Title: Real-time effectiveness evaluation of online education based on LSTM-transformer model

Authors: Changlin Wang

Addresses: School of Art, Jiujiang University, Jiujiang, 332000, China

Abstract: Online education, with its flexibility, has become an integral part of the education sector. To address the challenges posed by existing research, which struggles to capture spatio-temporal locality and handle lengthy historical evaluation sequences, this paper first inputs historical evaluation data into a long short-term memory network (LSTM) to discover long-term sequential relationships in the evaluation data. The LSTM's output is then fed into the Transformer encoder, followed by an encoding layer that feeds into the transformer layer, where multi-head attention mechanisms enhance concurrent learning of long-term dependencies. Second, the final evaluation prediction results are obtained through a softmax output. Finally, an improved Bayesian optimisation algorithm is used for hyperparameter iteration, and the optimal hyperparameters for the evaluation model are selected. Experimental outcome demonstrates that the average evaluation accuracy of the proposed model has improved by 5.98%-12.24%, validating the efficiency of the proposed model.

Keywords: online education; spatial layout; effectiveness evaluation; LSTM model; transformer model; Bayesian optimisation.

DOI: 10.1504/IJRIS.2025.150503

International Journal of Reasoning-based Intelligent Systems, 2025 Vol.17 No.12, pp.24 - 34

Received: 13 Jul 2025
Accepted: 12 Oct 2025
Published online: 15 Dec 2025 *

Title: Real-time effectiveness evaluation of online education based on LSTM-transformer model

Keep up-to-date