Scholarlink

<p>At present, College English reading teaching based on computer technology has the problem of insufficient discourse reasoning ability. This study proposes a deep learning framework, which combines the simple data expansion (EDA) with the transformer (BERT) model represented by the bidirectional encoder to build a text reasoning prediction model, so as to enhance the ability of academic English text reasoning. The English Wikipedia dataset is used to evaluate the performance of the model. The results show that the prediction error of the optimized model is only 0.67%±0.01%, the reasoning time is shortened to 2.1s ± 0.2s, and the prediction error rate is only 0.8%, which is better than the traditional natural language processing model. The model has been verified in College English reading text reasoning. The average reasoning time for English literature is 1.2&#xa0;min, and the similarity of text reasoning is 98.8%. The results show that this method can improve the understanding ability of academic English by improving discourse reasoning. This research not only provides a new technical solution for the teaching of academic English reading, but also can promote the application and development of deep learning model in the field of text reasoning, which is expected to significantly improve learners’ Academic English understanding ability, and have a far-reaching impact on the teaching and research in related fields.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing academic English text reasoning via EDA-optimized BERT

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月4日

查看全文
  • Jing Wu

摘要

At present, College English reading teaching based on computer technology has the problem of insufficient discourse reasoning ability. This study proposes a deep learning framework, which combines the simple data expansion (EDA) with the transformer (BERT) model represented by the bidirectional encoder to build a text reasoning prediction model, so as to enhance the ability of academic English text reasoning. The English Wikipedia dataset is used to evaluate the performance of the model. The results show that the prediction error of the optimized model is only 0.67%±0.01%, the reasoning time is shortened to 2.1s ± 0.2s, and the prediction error rate is only 0.8%, which is better than the traditional natural language processing model. The model has been verified in College English reading text reasoning. The average reasoning time for English literature is 1.2 min, and the similarity of text reasoning is 98.8%. The results show that this method can improve the understanding ability of academic English by improving discourse reasoning. This research not only provides a new technical solution for the teaching of academic English reading, but also can promote the application and development of deep learning model in the field of text reasoning, which is expected to significantly improve learners’ Academic English understanding ability, and have a far-reaching impact on the teaching and research in related fields.