情报科学 ›› 2025, Vol. 43 ›› Issue (7): 106-114.

• 业务研究 • 上一篇    下一篇

基于共现关系和语义相似关系的跨学科知识元对齐研究
——以图书情报学和管理科学为例

  

  • 出版日期:2025-07-05 发布日期:2025-10-16

  • Online:2025-07-05 Published:2025-10-16

摘要:

【目的/意义】为丰富学科领域知识体系,促进学科间知识交流,本文结合共现关系和语义相似关系,开展细
粒度的跨学科知识元对齐研究。【方法/过程】以图书情报学为目标学科,管理科学为源学科,借助SciAIEngine平台
和Python工具,抽取问题、方法知识元,建立“问题-方法”共现关系。在此基础上,通过语义相似性发现两学科间
潜在的知识关联,进而实现管理科学与图书情报学的跨学科知识元对齐。【结果/结论】研究发现了管理科学与图书
情报学对齐的多种方法和图书情报学未来能辅助解决的多项管理科学问题。并通过新颖性评估和可行性分析,验
证了跨学科知识元对齐效果,说明共现关系和语义相似关系结合的方法能够有效实现跨学科知识元的对齐。【创
新/局限】本研究从知识元视角实现了跨学科知识对齐,为交叉学科的科研创新提供了新思路。研究局限性在于,
借助SciAIEngine平台抽取知识元,人工干预度较高,未来将尝试基于深度学习的方法抽取知识元,以进一步提高知
识抽取的准确性。

Abstract:

【Purpose/significance】This study explores interdisciplinary knowledge alignment by combining co-occurrence and seman⁃
tic similarity, aiming to enrich academic knowledge systems and promote interdisciplinary knowledge exchange.【Method/process】Us⁃
ing Library and Information Science (LIS) as the target and Management Science (MS) as the source, the study employs SciAIEngine
and Python to extract problem-method knowledge elements and establish co-occurrence relationships. Semantic similarity is then
used to align knowledge elements between the two disciplines.【Result/conclusion】The study identifies methods in MS that align with
LIS and problems in MS that LIS can address. Novelty and feasibility analysis confirms the effectiveness of combining co-occurrence
and semantic similarity for interdisciplinary alignment.【Innovation/limitation】This study aligns interdisciplinary knowledge from the
perspective of knowledge elements, offers new insights for cross-disciplinary innovation. Although the SciAIEngine involves signifi⁃
cant manual intervention, future work will explore deep learning methods to improve extraction accuracy.