情报科学 ›› 2024, Vol. 42 ›› Issue (9): 123-134.

• 业务研究 • 上一篇    下一篇

基于知识图谱的数字记忆领域知识发现研究

  

  • 出版日期:2024-09-01 发布日期:2024-11-06

  • Online:2024-09-01 Published:2024-11-06

摘要: 【目的/意义】基于数字记忆研究的跨学科属性,本研究构建学术知识图谱,探索领域知识组织,挖掘潜在的 知识关联,以期为后续相关研究提供借鉴。【方法/过程】以图书馆学、情报学、档案学、计算机科学和新闻传播学学 科领域内发表的数字记忆研究学术论文为研究对象,通过主题建模,识别领域内的核心研究主题;借助SciAIEngine 和哈工大LTP 进行文本挖掘,抽取三元组构建领域学术知识图谱;通过知识查询可视化呈现的理论与方法,从“研 究理论”“研究方法”与“研究问题”视角分别实现知识发现。【结果/结论】基于所构建的知识图谱,归纳数字记忆领 域热门研究主题,开展了该领域研究内容的细粒度挖掘与聚合,为科学研究提供新视角。【创新/局限】研究对象关 注的学科领域范围有限,仅选择中文学术论文进行文本挖掘,未来应继续将其他相关学科纳入研究范围,同时补充 更多的外文文本。

Abstract: 【Purpose/significance】Based on the interdisciplinary nature of digital memory research, this study constructs academic knowledge map, explores domain knowledge organization, and excavates potential knowledge correlation, in order to provide reference for subsequent relevant research.【Method/process】Academic papers on digital memory published in the fields of library science, in⁃ formation science, archives science, computer science and journalism and communication science were taken as the research objects, and the core research topics in the fields were identified through theme modeling. With the help of SciAIEngine and LTP of HIT, text mining was carried out, and triples were extracted to construct the domain academic knowledge map. Through the theory and method of visual presentation of knowledge query, knowledge discovery is realized from the perspectives of "research theory", "research method" and subject research content.【Result/conclusion】Based on the constructed knowledge graph, the popular research topics in the field of digital memory were summarized, and the fine-grained mining and aggregation of the research contents in this field were carried out to provide a new perspective for scientific research.【Innovation/limitation】The subject area of the research object is limited, and only Chinese academic papers are selected for text mining. In the future, other relevant disciplines should continue to be included in the re⁃ search scope, and more foreign language research texts should be added.