情报科学 ›› 2023, Vol. 41 ›› Issue (10): 136-147.

• 博士论坛 • 上一篇    下一篇

面向唐诗作者风格画像的知识组织与验证

  

  • 出版日期:2023-10-01 发布日期:2023-12-05

  • Online:2023-10-01 Published:2023-12-05

摘要:

【目的/意义】唐诗作者风格要素知识组织是为智能实现唐诗作者风格画像提供知识服务的,是机器识别唐
诗作者身份的重要突破,能够为数字人文研究文献辑佚和文化传播提供支撑。【方法/过程】本文首先梳理了语言风
格学相关理论和传统人文领域专家学者们关于唐诗风格的品鉴方法和策略,以及数字人文唐诗研究形成的语料库
和检索平台;其次,根据专家提供的领域知识和唐诗文本自身的特点,构建了唐诗风格要素系统框架;在此基础上,
面向唐诗作者风格画像需求,从作者基本属性要素知识、文体风格特征要素知识和价值属性要素知识三个维度进
行了详细分析,给出了不同层面的知识组织。【结果/结论】通过选取知识组织中的要素知识进行了适用性验证,结
果表明,不同维度的要素知识能够实现相应画像侧写。【创新/局限】研究发现:唐诗作为古文献进行数字化研究涉
及的理论和领域知识较多,以及对机器识别与人工识别的区别认识不足等因素,造成研究具有一定的局限性。

Abstract:

【Purpose/significance】 The knowledge organization of tang poetry author style elements provides knowledge services for the
intelligent implementation for poets profiling in Tang Dynasty, which is an important breakthrough in authorship attribution for poetry in Tang Dynasty. It can provide support for the literature collection in digital humanities and cultural dissemination.【
Method/process】In this paper, we first reviewed the relevant theories of linguostylistics and summarized the evaluation methods and strategies on the style of Tang poetry provided by traditional humanities experts and scholars. We also reported the corpus and retrieval platform formed by digital humanities research on Tang poetry. Secondly, based on the domain knowledge provided by experts and particularity of Tang poetry, we established a systematic framework of Tang poetry style elements. In order to satisfied the demand of poets profiling, a de⁃tailed analysis was conducted from three dimensions: the authorship attribution element knowledge, stylistic feature element knowl⁃edge, and value attribution element knowledge, and provide a knowledge organization in different levels. 【Result/conclusion】 Applica⁃bility verification was conducted by selecting element knowledge from the knowledge organization, and the results showed that differ⁃ent dimensions of element knowledge can achieve corresponding poets profiling.【Innovation/limitation】 Research shows that the digi⁃tization of Tang poetry as ancient literature involves a large amount of theoretical and domain knowledge. However, we hardly recog⁃nized the difference between machine recognition and human recognition, which has led to a certain limitation in the research.