摘要
在古籍数字化时代背景下,针对文学类古籍实体重名等问题,探究如何基于多元关系表达模式构建知识图谱。在梳理古籍知识服务及知识问答系统相关研究的基础上,基于自然语言技术与模块匹配提出融合多元知识关联的文学类古籍问答系统架构,优化现有抽取模型与基于知识图谱的问答系统构建框架,多维度挖掘其人文性知识,以多元关系表达模式存储数据,选取《聊斋志异》进行实证研究。实验结果表明经设计优化后的文学类古籍问答系统在知识推理与智能交互方面表现良好,意图识别与知识推理能力得到有效提升,基本满足交互性需求,拓宽了文学类古籍知识图谱应用服务维度,为今后文学类古籍领域数字化传承与智能化建设奠定基础。
Under the background of the digital age of ancient books,this paper explores how to construct knowledge graph based on multiple relation expressions,aiming at the problem of entity name of ancient literature books.On the basis of sorting out relevant studies on ancient books knowledge service and knowledge question answering system,based on natural language technology and module matching,this paper proposes the architecture of ancient literature books question answering system integrating multiple knowledge associations.Optimize the existing extraction model and the framework for building a question-answering system based on knowledge graphs,the humanistic knowledge was mined in multiple dimensions,the data was stored in the multi-relational expression mode,Liaozhai Zhiyi was selected for empirical study.The experimental results show that the optimized question answering system of ancient literature books performs well in knowledge reasoning and intelligent interaction,the ability of intention recognition and knowledge reasoning is effectively improved,basically meet the demand for interaction,and broaden the application service dimension of knowledge graph of ancient literature books.It lays a foundation for the digital inheritance and intelligent construction of ancient literature books in the future.
作者
王克平
范颜铄
许明珠
周婧怡
Wang Keping;Fan Yanshuo;Xu Mingzhu;Zhou Jingyi
出处
《新世纪图书馆》
2025年第3期57-66,共10页
New Century Library
基金
国家社会科学基金项目“基于‘互联网+’思维的新创企业风险识别与竞争情报预警研究”(项目编号:20BTQ052)系列研究成果之一。
关键词
文学类古籍
知识组织
自然语言处理
问答系统
Ancient literature books
Knowledge organization
Natural language processing
Question-answering system
