Exploring the Feasibility of Large Language Model- and Rubric-Based Automatic Assessment of Elementary Students’ Book Summaries

Qi-Zhen Huang, Hou-Chiang Tseng, Yao-Ting Sung


Abstract
摘要寫作為閱讀與寫作整合的高層次語文任務,不僅可評量學生的文本理解能力,也能促進語言表達與重述能力的培養。過去自動摘要批改系統多依賴關鍵詞比對或語義重疊等「由下而上」的方法,較難以全面評估學生的理解深度與文本重述能力,且中文摘要寫作批改研究雖有,但相較於英文仍相對不足,形成研究缺口。隨著大型語言模型(Large Language Models, LLMs)的發展,其在語意理解與生成能力上的突破,為自動摘要批改與回饋帶來新契機。有鑑於此,本研究旨以由上而下的方式探討結合LLMs與閱讀摘要評分規準(Rubrics)對學生閱讀摘要批改與回饋之應用潛力,進一步而言,在考量教學資料隱私的情況下,本研究採用Meta-Llama-3.1-70B生成電腦摘要,並依據專家所制定的摘要評分規準,其評分涵蓋:理解與準確性、組織結構、簡潔性、語言表達與文法及重述能力五大構面,對學生閱讀摘要進行自動評分與回饋。研究結果顯示,Meta-Llama-3.1-70B能提供具體、清晰的即時回饋,不僅能指出摘要中遺漏的關鍵概念,也能針對結構安排與語法錯誤提出修正建議,協助學生快速掌握摘要改進方向;然而回饋多偏向表面語言與結構調整,在語言表達、修辭多樣性及重述能力等高層次語文能力評估上仍存在限制。整體而言,LLMs可作為形成性評量與教學輔助工具,提升評分效率,但需結合教師專業判斷與回饋以補足深層概念與策略性寫作指導,促進學生摘要寫作能力的發展。
Anthology ID:
2025.rocling-main.19
Volume:
Proceedings of the 37th Conference on Computational Linguistics and Speech Processing (ROCLING 2025)
Month:
November
Year:
2025
Address:
National Taiwan University, Taipei City, Taiwan
Editors:
Kai-Wei Chang, Ke-Han Lu, Chih-Kai Yang, Zhi-Rui Tam, Wen-Yu Chang, Chung-Che Wang
Venue:
ROCLING
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
153–166
Language:
URL:
https://aclanthology.org/2025.rocling-main.19/
DOI:
Bibkey:
Cite (ACL):
Qi-Zhen Huang, Hou-Chiang Tseng, and Yao-Ting Sung. 2025. Exploring the Feasibility of Large Language Model- and Rubric-Based Automatic Assessment of Elementary Students’ Book Summaries. In Proceedings of the 37th Conference on Computational Linguistics and Speech Processing (ROCLING 2025), pages 153–166, National Taiwan University, Taipei City, Taiwan. Association for Computational Linguistics.
Cite (Informal):
Exploring the Feasibility of Large Language Model- and Rubric-Based Automatic Assessment of Elementary Students’ Book Summaries (Huang et al., ROCLING 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.rocling-main.19.pdf