Applying the BERT Model Combined with Multi-Task Learning to Improve the Accuracy of Academic Automatic Scoring in Higher Education Paper Writing

DOI

10.2991/978-2-38476-452-5_14How to use a DOI?

Keywords

BERT Model; Academic Scoring; Contextual Semantic Understanding; Educational Artificial Intelligence; Multi-Task Learning

Abstract

The current automatic scoring of academic content in higher education paper writing has problems such as insufficient accuracy and inconsistent scoring standards. To address these challenges, this paper proposes an innovative method that combines the BERT model and multi-task learning. Specifically, this paper uses the pre-trained BERT (Bidirectional Encoder Representations from Transformers) model as the basis for text representation, making full use of its powerful contextual semantic understanding ability to accurately analyze the academic content of papers. Next, in view of the multiple dimensions involved in academic scoring, this paper designs a multi-task learning framework that allows the model to share knowledge between different tasks, thereby improving the performance of the model in scoring in each dimension. In order to further enhance the robustness of the model, this paper also adds a data augmentation strategy during the training process, using multiple manually annotated scoring samples for joint training. The experimental results show that the accuracy of the model in terms of logic scoring is 84.1%, and in terms of language quality, it is 87.5%. The experiment also verifies that the method has strong generalization ability in cross-domain adaptability, especially when dealing with papers from different disciplines, it is more stable and accurate. In summary, the method proposed in this paper provides a more accurate and consistent solution for academic automatic scoring in higher education paper writing.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Danni Xie
PY  - 2025
DA  - 2025/07/31
TI  - Applying the BERT Model Combined with Multi-Task Learning to Improve the Accuracy of Academic Automatic Scoring in Higher Education Paper Writing
BT  - Proceedings of the 2025 4th International Conference on Science Education and Art Appreciation (SEAA 2025)
PB  - Atlantis Press
SP  - 107
EP  - 116
SN  - 2352-5398
UR  - https://doi.org/10.2991/978-2-38476-452-5_14
DO  - 10.2991/978-2-38476-452-5_14
ID  - Xie2025
ER  -