论文部分内容阅读
【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。
[Purpose / Significance] Through the multi-modal analysis of heterogeneous networks, more topics can be integrated to improve the accuracy of cross-subject identification. Methods / Processes The existing methods of community identification of 2-mode heterogeneous networks are systematically investigated. The direct identification of 2-mode network communities is divided into projection methods, non-projection methods, extended multimode network analysis and super-network analysis. And compares the advantages and disadvantages of these existing methods in the subject cross-subject identification. On the basis of this, we choose correspondence analysis as the method of 2-modal network community identification to identify and analyze the subject of information science. [Results / Conclusions] Correspondence analysis as a kind of 2-modal relational analysis method can directly identify the subject of cross-disciplinary heterogeneous networks and ensure less information is missing.