| 张为,肖巧玲,刘海江,任好,蔡子妍,苏鹏鹍,顾进广.大模型辅助的汉语文化负载词抽取与知识图谱构建[J].数字图书馆论坛,2025,21(1):33~45 |
| 大模型辅助的汉语文化负载词抽取与知识图谱构建 |
| Large Model-Assisted Extraction and Knowledge Graph Construction of Chinese Culture-Loaded Words |
| 投稿时间：2024-10-27 |
| DOI：10.3772/j.issn.1673-2286.2025.01.005 |
| 中文关键词: 大语言模型；知识图谱构建；文化负载词；知识抽取；文本分类 |
| 英文关键词: Large Language Model; Knowledge Graph Construction; Culture-Loaded Word; Knowledge Extraction; Text Classification |
| 基金项目:本研究得到国家重点研发计划重点专项（编号：2022YFC3300801）、湖北省教育厅科学研究计划指导项目（编号：B2019009）、武汉市
知识创新专项-曙光计划项目（编号：2023010201020409）、文化负载词知识库及其在跨文化领域示范应用（编号：24Z346）资助。 |
| 作者 | 单位 | | 张为 | 武汉科技大学计算机科学与技术学院；湖北省智能信息处理与实时工业系统重点实验室 | | 肖巧玲 | 武汉科技大学国际教育学院 | | 刘海江 | 武汉科技大学计算机科学与技术学院；湖北省智能信息处理与实时工业系统重点实验室 | | 任好 | 武汉科技大学计算机科学与技术学院 | | 蔡子妍 | 武汉科技大学计算机科学与技术学院 | | 苏鹏鹍 | 武汉科技大学计算机科学与技术学院 | | 顾进广 | 武汉科技大学计算机科学与技术学院；湖北省智能信息处理与实时工业系统重点实验室 |
|
| 摘要点击次数: 27 |
| 全文下载次数: 56 |
| 中文摘要: |
| 文化负载词能够反映民族文化和风俗，是文化认同与情感共鸣的纽带，广泛应用于跨文化交流、教育等领域。目前研究主要集中于负载词的翻译和教学，缺乏对词汇本身的规范化定义和描述。知识图谱具有整合信息和语义关联的优势，应用于文化负载词领域可推动其规范化组织，促进其在翻译、教育等方面的应用。因此，综合现有对文化负载词分类的研究，设计可扩展的汉语文化负载词本体模型，并从文献、教材以及权威网站收集相关的中文语料，利用UIE模型进行知识抽取。为描述各负载词之间的关联，使用大语言模型，根据上下文对汉语文化负载词进行层次分类以及关系补全。结果表明，基于提出的层次化可扩展本体模型构建的知识图谱，能够系统整合汉语文化负载词的相关属性，映射词汇间的文化关联，解决了传统知识图谱在处理多义词和动态表达时存在的问题，为翻译和文化教育等领域提供准确的语义支持，从而提高文化知识的解释性和实际应用效果。 |
| 英文摘要: |
| Culture-loaded words reflect national culture and customs, serving as a bridge for cultural identity and emotional resonance. They are widely used in areas such as cross-cultural communication and education. Current research mainly focuses on the translation and teaching of culture-loaded words, lacking standardized definitions and descriptions of the words themselves. Knowledge graphs, with their advantages in information integration and semantic relationships, can promote the standardized organization of culture-loaded words and facilitate their application in translation, education, and other fields. Therefore, this paper synthesizes existing research on the classification of culture-loaded words and designs an expandable Chinese culture-loaded word ontology model. Relevant Chinese corpus is collected from literature, textbooks, and authoritative websites, and knowledge extraction is performed using the UIE model. To describe the relationships between various culture-loaded words, this paper employs large language models to perform hierarchical classification and relationship completion based on context. The results indicate that the knowledge graph constructed based on the hierarchical extensible ontology model proposed in this article can systematically integrate the relevant attributes of Chinese culture-loaded words, map the cultural associations between vocabulary, solve the problems of traditional knowledge graphs in dealing with polysemous words and dynamic expressions, provide accurate semantic support for translation and cultural education fields, and improve the interpretability and practical application effect of cultural knowledge. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
