End-to-End Architecture for English Reading and Writing Content Assessment Based on Prompt Learning

• Published in: IEEE access 2024-11, p.1-1
• Main Author: Ji, Yun-Peng
• Format: Article
• Language: English
• Subjects: Accuracy; Adaptation models; Attention mechanisms; Automated Assessment; Context modeling; Deep Learning; End-to-End Model; Feature extraction; Few shot learning; Manuals; Multi-Scale Attention Mechanisms; Prompt Learning; Semantics; Text detection; Writing
• ISSN: 2169-3536
• DOI: 10.1109/ACCESS.2024.3509990

With the acceleration of globalization, the use of English as an international language has become increasingly widespread, making the assessment of English reading and writing skills a key issue in the field of language education. However, traditional methods of English reading and writing content assessment rely on manually designed features, which struggle to effectively handle complex and diverse language structures. To address this issue, this study proposes an English reading and writing content assessment algorithm based on Prompt Learning, utilizing an end-to-end architecture enhanced with multi-scale attention mechanisms. The algorithm preprocesses the input English texts through the prompt learning framework and uses multi-scale attention mechanisms to improve the model's ability to capture features at different linguistic levels. Within an end-to-end architecture, the entire assessment process is automated, from text input to output of assessment results, eliminating the need for manually designed feature extraction steps. Experimental results show that the algorithm performs excellently on multiple English reading and writing assessment datasets, significantly enhancing the accuracy and efficiency of assessments and offering an effective solution for English reading and writing evaluations.