Was Poetry Graded Validly?: Text Mining Shipin , a Sixth-CenturyChinese Work of Literary Criticism

2025/11/17 by Wenyi Shang, Emily Xueyue Liu · 1 voice
Computer Science · #Authorship Attribution and Profiling #Sentiment Analysis and Opinion Mining #Artificial Intelligence in Games

paper · pdf · doi:10.63744/unuzr0wn2vsq

Abstract

This paper applies text mining to investigate <em>Shipin</em> (Poetry Gradings), a sixth-century Chinese work of literary criticism. Using a BERT model fine-tuned with masked language modeling on a classical Chinese poetry corpus, we generated embeddings for <em>Shipin</em>’s evaluative remarks on each poet and their own poetry corpora, and explored the relationship between these embeddings and the grades assigned to each poet by <em>Shipin</em> using PCA and machine learning classification. We found that both remarks and poetry provide some justification for the assigned grades, with remarks showing a much closer alignment. A poet’s dynastic period and poetic origin influenced the grades they received in nuanced ways, reflecting <em>Shipin</em>’s preference for poetic styles. The results indicate that <em>Shipin</em> maintained an implicit but consistent standard in grading.

Discussions

• Excited to present my work with @emilyue.bsky.social at #CHR2025 tomorrow at 3pm! If you're around, please join us. Our full paper is available at: doi.org/10.63744/uNU.... We fine-tuned a BERT model [bsky, 13 points, 2 comments]

Related

• Understanding Literary Texts by LLMs: A Case Study of Ancient Chinese Poetry
• Capabilities and Evaluation Biases of Large Language Models in Classical Chinese Poetry Generation: A Case Study on Tang Poetry
• Benchmarking the Detection of LLMs-Generated Modern Chinese Poetry
• Generation of Chinese classical poetry based on pre-trained model
• CCPM: A Chinese Classical Poetry Matching Dataset