Abstract
本文实验发现,段落顺序会影响人类阅读理解效果;而打乱段落或句子顺序,对BERT、ALBERT和RoBERTa三种人工神经网络模型的阅读理解答题几乎没有影响。打乱词序后,人的阅读理解水平低于三个模型,但人和模型的答题情况高于随机水平,这说明人比人工神经网络对词序更敏感,但人与模型可以在单词乱序的情况下答题。综上,人与人工神经网络在正常阅读的情况下回答阅读理解问题的正确率相当,但两者对篇章结构及语序的依赖程度不同。- Anthology ID:
- 2021.ccl-1.41
- Volume:
- Proceedings of the 20th Chinese National Conference on Computational Linguistics
- Month:
- August
- Year:
- 2021
- Address:
- Huhhot, China
- Editors:
- Sheng Li (李生), Maosong Sun (孙茂松), Yang Liu (刘洋), Hua Wu (吴华), Kang Liu (刘康), Wanxiang Che (车万翔), Shizhu He (何世柱), Gaoqi Rao (饶高琦)
- Venue:
- CCL
- SIG:
- Publisher:
- Chinese Information Processing Society of China
- Note:
- Pages:
- 446–453
- Language:
- Chinese
- URL:
- https://aclanthology.org/2021.ccl-1.41
- DOI:
- Cite (ACL):
- Shukai Ma, Jiajie Zou, and Nai Ding. 2021. 基于篇章结构攻击的阅读理解任务探究(Analysis of Reading Comprehension Tasks based on passage structure attacks). In Proceedings of the 20th Chinese National Conference on Computational Linguistics, pages 446–453, Huhhot, China. Chinese Information Processing Society of China.
- Cite (Informal):
- 基于篇章结构攻击的阅读理解任务探究(Analysis of Reading Comprehension Tasks based on passage structure attacks) (Ma et al., CCL 2021)
- PDF:
- https://preview.aclanthology.org/emnlp-22-attachments/2021.ccl-1.41.pdf
- Data
- RACE