数字人文视角下古代农书语义知识组织与多维知识发现研究——以“四大农书”为对象
y. wu1, q. zhang2, s. zhou3
1华中师范大学, China, Central China Normal University; 2华中师范大学, China, Central China Normal University; 3华中师范大学, China, Central China Normal University
[目的/意义]古农书是不可再生的文化资源，记载了中华传统农耕文化，对其史料资源进行知识组织与知识发现可以推进古农书馆藏文献的数字化开发进程，有助于实现对古农书史料资源的深度开发与智慧服务。[方法/过程]本研究以“古代四大农书”为研究对象，构建了基于知识图谱的古农书资源语义知识组织框架。首先，通过对数据进行分析与整理，构建古农书史料资源本体模型实现对资源中的概念与关系的规范化描述，其次，基于已构建的模型进行知识抽取，利用知识融合技术来消除共指问题，最后将RDF数据存储到Neo4j图数据库中，实现对古农书史料资源的语义知识组织与多维知识发现。[结果/结论]本研究为古农书史料资源的语义知识组织与多维知识发现提供了研究方案，也为数字人文背景下古农书资源深层次开发与利用提供了新的研究视角。
基于ChatGPT+Prompt的中国传统戏剧类非物质文化遗产知识图谱构建研究
S. Zhou1,3, H. Chen1, J. Zhang2,3
1School of Information Management, Central China Normal University, Wuhan, China, People's Republic of; 2Sichuan University of Culture and Arts, Mianyang, China, People's Republic of; 3Digital Media Art, Key Laboratory of Sichuan Province, Sichuan Conservatory of Music, Chengdu, China, People's Republic of
中国传统戏剧类非遗是我国表演艺术的瑰宝，是中华民族智慧的结晶，也是广大人民群众喜闻乐见的艺术表现形式，承载着丰富的历史、艺术和社会内涵。随着数字技术的不断进步，现代技术为非物质文化遗产的数字化、知识化、智慧化转变提供了新的机遇，细粒度的语义描述和关联组织成为领域研究所需。本研究旨在借助ChatGPT+Prompt模式构建中国传统戏剧类非物质文化遗产知识图谱，从而促进对这一珍贵文化遗产的更深入理解和传承。首先，以中国传统戏剧类非遗为研究对象，通过爬虫技术获取原始文本数据集，设计了CRISPE架构的问答提示语（Prompt）来引导知识图谱构建的全过程。其次，基于大语言模型的ChatGPT，采用对话式的few-shot学习方式，设计了一套针对戏剧类非遗的知识本体，实现了知识的有效迁移。其次，通过ChatGPT进行知识抽取，成功识别出非物质文化遗产的实体和关系，并将其存储至图数据库Neo4j中，实现了Text2SOL的过程。结果表明，通过本文的研究不仅有助于数字化保护和传承中国传统戏剧类非物质文化遗产，还为构建领域知识图谱的新方法提供了有益的启示。通过结合大语言模型和Prompt技术，为非物质文化遗产的数字化管理和推广开辟了新的途径，提升了其社会和文化价值，同时也有望为其他领域引入大语言模型提供了新的思路。
The evolution of knowledge on sericulture and mulberry technology based on agricultural ancient books
W. Wang, Y. Sun, S. Xiong, Z. Zhang
School of Information Management, Wuhan University, Wuhan, 430072, China;
[Purpose/significance] By focusing on agricultural texts to analyse the evolution of ancient scientific knowledge, we can discover the development of ancient science and technology and pass on the excellent traditional science and technology. [Method/process] The article first constructs a corpus of agricultural book texts based on sericulture book records, extracts keywords from the texts using the tf-idf algorithm, builds a co-occurrence matrix using the co-occurrence relationship of keywords, and then constructs a knowledge network of agricultural books; adopts the Louvain community detection algorithm for community division and topic discovery of the knowledge network; determines the knowledge community by calculating the community similarity of adjacent time evolution and visualize the presentation. [Results/conclusion] The research has discovered that the knowledge of mulberry cultivation and silk reeling was inherited stably during the Yuan, Ming, and Qing dynasties, which has enhanced our understanding of the knowledge system of sericulture history. It has also revealed other significant themes in sericulture history, particularly the importance of early silkworms in the field of sericulture technology and the consistent transmission of silkworm breeding themes within the sericulture knowledge cluster. Additionally, the research has uncovered the developmental features of historical sericulture technology knowledge, with a continuous accumulation of sericulture knowledge, reaching its peak during the Qing dynasty.
三元空间视域下重大突发事件的事件图谱研究
刘. 伟利1, 张. 海涛1,2,3, 张. 鑫蕊1, 周. 红磊1
1吉林大学商学与管理学院; 2吉林大学信息资源研究中心; 3吉林大学国家发展与安全研究院
摘要：[目的/意义]提出融合三元空间大数据的重大突发事件的事件图谱模型，旨在支撑重大突发事件的全要素、全过程、全方位知识表示。[方法/过程]首先，从世界图景的概念出发界定事件图景的内涵和类型，并构建通用事件图谱模型，用以关联不同抽象层次的知识，承载多维立体的事件图景；其次，探究重大突发事件在物理-社会-网络三元空间演化过程，借助事物运动与信息传播两个过程，揭示重大突发事件的跨空间演化机理，构建重大突发事件的跨空间演化模型；最后，基于重大突发事件的跨空间演化模型构建重大突发事件的事件图谱模型。[结果/结论] 知识表示方面，重大突发事件的事件图谱不仅可以描述事件在三元空间的演化过程，揭示重大突发事件的演化模式，还可以呈现物理、社会和网络空间要素的状态及其间相互作用，揭示事件演化的内在动力。知识组织方面，重大突发事件的事件图谱能够融合三元空间的关键信息，关联碎片化知识，呈现一个立体、多维的事件图景。
Personalized Recommendation Rationality on Content Platforms: Construct and Effect
G. Li, M. Wang
Sichuan University, China, People's Republic of
Personalized recommendations have been widely employed in the commercial development of various content platforms, and users' awareness of them has deepened. While there is already a substantial body of research on the evaluation of recommendation algorithms from a user-centric perspective, there is a lack of research evaluating personalized recommendation algorithms from an ethical dimension. Therefore, this study, from the perspectives of both users and public interest, applies the normative analysis method to propose a new concept, 'Rationality of Personalized Recommendations on Content Platforms,' along with its three dimensions: Technical Rationality, Content Rationality, and Ethical Rationality. Building upon existing literature, this study formulates operational indicators and constructs a scale with the aim of confirming a theoretical model for constructing the impact of the rationality of recommendations on users' intention to continue use. An empirical research method is employed to explore the causal mechanisms. The results indicate that the newly developed scale for the rationality of personalized recommendations on content platforms exhibits high reliability and validity. Furthermore, it demonstrates a significant positive impact on users' intention to continue use.
|
