CLGC: A Corpus for Chinese Literary Grace Evaluation

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 5548–5556 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 5548 CLGC: A Corpus for Chinese Literary Grace Evaluation Yi Li, Dong Yu� , Pengyuan Liu School of Information Science, Beijing Language and Culture University Beijing, China liyi blcu@163.com, yudong blcu@126.com, liupengyuan@pku.edu.cn Abstract In this paper, we construct a Chinese literary grace corpus, CLGC, with 10,000 texts and more than 1.85 million tokens. Multi-level annotations are provided for each text in our corpus, including literary grace level, sentence category, and figure- of-speech type. Based on the corpus, we dig deep into the correlation between fine-grained features (semantic information, part-of-speech and figure-of-speech, etc.) and literary grace level. We also propose a new Literary Grace Evaluation (LGE) task, which aims at making a comprehensive assessment of the literary grace level according to the text. In the end, we build some classification models with machine learning algorithms (such as SVM, TextCNN) to prove the effectiveness of our features and corpus for LGE. The results of our preliminary classification experiments have achieved 79.71% on the weighted average F1-score. Keywords: corpus, annotation, Chinese literary grace, linguistic features 1. Introduction Literary grace (文采) reveals the aesthetic function of language and matters of the use by a writer of the lan- guage (Wang, 1994; Rastall, 2008). Texts with better literary grace can bring readers a higher aesthetic ex- perience. For instance, when reading the two exam- ple sentences in Table 1, we can obviously feel the higher aesthetic values of the first sentence. Literary grace is embodied in phonetics, form, and semantic, including beauties of sound, modification, color, emo- tion, image, and philosophy. It is regarded as one of the elements constituting the style and quality of arti- cles in China (Wang, 1994; Zhou, 2013; Qi, 2006). The discussion about it has been throughout all the ancient Chinese literary theories. As an indispensable factor of the text, literary grace also has attracted scholars in the natural language processing (NLP) field. Many studies have referred to several aspects of it and been applied in real-world applications, including Automated Essay Evaluation (AEE) (Liu et al., 2016a), quality assess- ment (Qiao et al., 2021), machine translation, opinion mining, dialogue modeling, and modeling argumenta- tive discourse (Tong et al., 2021). However, there have been few publicly available cor- pora annotated with literary grace. The traditional re- lated literature and linguistics study is based on the scattered examples or limited to a specific text, not forming a large-scale and integrated corpus. The ex- isting task-oriented datasets are only focused on the sub aspects of literary grace: figure-of-speech (Birke and Sarkar, 2006; Liu et al., 2018) and rhetoric level (Shi, 2019; Gong, 2016). Both of them are ways to get aesthetic feelings, not equivalent to literary grace. More recently, Fu et al. (2018) constructed a sentence- level dataset that is closely related to literary grace. However, the generation of literary grace is also af- � Corresponding …

Content truncated to 500 words