2025/06/05 by Léane Jourdan, Jourdan, Léane, Florian Boudin +5 · 1 citation
Arts and Humanities · Computer Science · #Academic Writing and Publishing #Computation and Language (cs.CL) #FOS: Computer and information sciences #Text Readability and Simplification #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2506.04772
openalex publication_date 2025/06/05 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Evaluating text revision in scientific writing remains a challenge, as traditional metrics such as ROUGE and BERTScore primarily focus on similarity rather than capturing meaningful improvements. In this work, we analyse and identify the limitations of these metrics and explore alternative evaluation methods that better align with human judgments. We first conduct a manual annotation study to assess the quality of different revisions. Then, we investigate reference-free evaluation metrics from related NLP domains. Additionally, we examine LLM-as-a-judge approaches, analysing their ability to assess revisions with and without a gold reference. Our results show that LLMs effectively assess instruction-following but struggle with correctness, while domain-specific metrics provide complementary insights. We find that a hybrid approach combining LLM-as-a-judge evaluation and task-specific metrics offers the most reliable assessment of revision quality.