classical-chinese

本文以基于最小对立的诊断框架，研究机器翻译的自动评估指标是否适用于文言文到英语的翻译。结果发现所有指标都存在盲点，其中 MetricX24 整体表现最佳。

This paper presents a computational framework for automatically compiling collected commentaries on classical Chinese texts, preserving contextual dependencies of inline notes via prompt chaining and cross-source clustering.

This paper recasts fine-grained intertextuality extraction in Classical Chinese histories as an agentic LLM task, grounding reuse in exact character spans and a five-dimension typology, validated by expert-adjudicated benchmarks and scaled to the Twenty-Four Histories.

本文介绍Lepton，一个微调的BERT分类器，用于预测古典中文文集目录中的标题是私人信件还是序言（尤其是易混淆的赠序）。该模型利用了来自明末清初三十三位文人的5,438条手工标注标题。