基于异构网络的学科交叉主题发现方法

发布时间：2018-05-10 19:51

本文选题：学科交叉 + 多模网路　；参考：《情报科学》2017年06期

【摘要】：【目的/意义】通过异构网络的多模关系分析可以整合更多的主题关联关系,从而提高学科交叉主题识别的准确度。【方法/过程】系统调研了已有的2-模异构网络的社区识别方法,将2-模网络社区直接识别分为投影方法、非投影方法、扩展的多模网络分析和超网络分析,并对比这些已有方法在学科交叉主题识别中的优劣。在此基础上选择对应分析作为2-模网络社区识别方法,对情报学的学科交叉主题进行识别和分析。【结果/结论】对应分析作为一种2-模关系分析方法可直接识别异构网络的学科交叉主题,并保证更少的信息遗漏。
[Abstract]:[purpose / meaning] more thematic relationships can be integrated through multimode analysis of heterogeneous networks, In order to improve the accuracy of interdisciplinary subject recognition. [method / process] the existing community recognition methods of 2-mode heterogeneous network are systematically investigated, and the community recognition of 2-mode network is divided into projection method and non-projection method. The extended multimode network analysis and supernetwork analysis are compared, and the advantages and disadvantages of these existing methods in interdisciplinary subject recognition are compared. On this basis, the correspondence analysis is selected as the community identification method of 2-mode network. This paper identifies and analyzes the subject intersection of information science. [results / conclusions] as a 2-norm relationship analysis method, the corresponding analysis can directly identify the interdisciplinary topics of heterogeneous networks and ensure less information omission.
【作者单位】：中国科学院成都文献情报中心;中国科学院大学;
【基金】：国家社会科学青年项目(14CTQ033) 中国科学院青年创新促进会资助
【分类号】：G350

【相似文献】