| 任东海,刘忠宝,张博文.数字人文视域下文学作品风格计量分析与比较研究—— 以金庸古龙小说为例[J].情报工程,2025,11(4):050-064 |
| 数字人文视域下文学作品风格计量分析与比较研究—— 以金庸古龙小说为例 |
| A Quantitative Analysis and Comparative Study on the Styles of Literary Works from the Perspective of Digital Humanities: A Case Study of Jin Yong’s and Gu Long’s Novels |
| |
| DOI： |
| 中文关键词: 作品风格分析；指标测度体系；机器学习；假设检验；数字人文 |
| 英文关键词: Literary Style Analysis; Metric System; Machine Learning; Hypothesis Testing; Digital Humanities |
| 基金项目:研究阐释党的二十大精神国家社科基金重点项目“大数据时代古籍活化赋能文化自信自强的理论、方法与路径研究”(23AZD047)。 |
| 作者 | 单位 | | 任东海 | 1.北京语言大学信息科学学院 北京 100083；2.河南经贸职业学院人工智能学院 郑州 450000 | | 刘忠宝 | 北京语言大学信息科学学院 北京 100083 | | 张博文 | 北京语言大学信息科学学院 北京 100083 |
|
| 摘要点击次数: 26 |
| 全文下载次数: 12 |
| 中文摘要: |
| [目的/意义]解决文学作品风格计量分析中缺乏系统性指标体系、指标测度方法适用性差等问题，为数字人文视域下文学作品风格的科学计量分析提供支撑。[方法/过程]首先，基于文学作品内容构成粒度，从微观到宏观选取词汇、句子、段落、篇章4 个层面的10 项核心指标，构建两级指标体系；其次，对指标测度方法进行优化，包括采用变异系数替代标准差测度离散度以解决均值偏差问题，定义词频熵并结合外部词频表测度作品通俗性，设计分组匹配法对比作品词汇丰富度以控制篇幅干扰；最后，基于金庸古龙武侠小说语料库，通过统计分析与机器学习验证该指标体系的有效性。[局限]当前研究指标体系仅以文本语言特征为主，未来可考虑整合情节网络、人物关系等非语言特征，构建全景式文学作品风格分析指标体系。[结果/结论]研究发现古龙与金庸作品风格存在多方面显著差异，且发现了古龙创作风格的演变规律，证实了所构建的指标体系在风格分析、作者识别与作品聚类中的应用有效性。 |
| 英文摘要: |
| [Objective/Significance] This study aims to address issues such as the lack of a systematic indicator system and poor applicability of indicator measurement methods in the quantitative analysis of literary work styles, providing support for the scientific quantitative analysis of literary work styles from the perspective of digital humanities. [Methods/Processes] Firstly,based on the content composition granularity of literary works, ten core indicators at four levels (vocabulary, sentence, paragraph,and text) from micro to macro are selected to construct a two-level indicator system. Secondly, the indicator measurement methods are optimized, including using the coefficient of variation instead of the standard deviation to measure dispersion to solve the problem of mean deviation; defining word frequency entropy and combining it with external word frequency tables to measure the popularity of works; and designing a group matching method to compare the vocabulary richness of works to control the interference of length. Finally, based on the corpus of Jin Yong’s and Gu Long’s martial arts novels, the effectiveness of the indicator system is verified through statistical analysis and machine learning. [Limitations] The current research indicator system only focuses on text language features. In the future, non-language features such as plot networks and character relationships can be integrated to construct a panoramic indicator system for literary work style analysis. [Results/Conclusions]The study found that there are significant differences in the styles of Gu Long’s and Jin Yong’s works in multiple aspects, and discovered the evolution law of Gu Long’s creative style, confirming the application effectiveness of the constructed indicator system in style analysis, author identification, and work clustering. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
