Page 198 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 44

197
Extended English abstracts of articles published in the Chinese edition of Journal of Library Science in China 2018 Vol.44 197

In addition, through the comparison of qualified papers and double-highly selected papers, it
is found that there are some differences between qualitative analysis using qualified papers as
references and quantitative analysis. The personalization analysis of 8 intersection papers shows
that the quality analysis of papers is a very complex and difficult problem, which cannot be
completely replaced by quantitative evaluation.
Finally, by comparing the domestic “download” research with the international “usage” study on
different datasets, we find that there are similarities between the two indexes in different datasets,
so that these similar conclusions should be universal.

Exploring the vital node pairs in information networks by deletion algorithm

ZHAO Xing〇
〇a*
The flow of information is one of the major values of information. In the evolving information
world, nodes are generated by information content and carrier. Meanwhile, links derive from
information behaviors and interactions. Then nodes and links construct networks. Such networks
present the abstract picture of the real information world.
In the previous studies of networks measures, the power-law distribution of nodes in complex
networks, the centrality measures in social networks, and h-type networks measures in Library
Science all focus on nodes instead of links. However, studying the interactions and relations of
information is an important task of information analysis. From the point of view of connection,
the most basic components can be expressed as node pairs, a combination of a link and nodes at
both ends. In information networks, a node pair usually represents a complete channel, which has
significant mining and measuring values.
This paper selects node pairs, which are the most basic component of links, as research object
and attempts to measure them indirectly. The study will use the “deletion algorithm”, which is
common in computer science, to construct the measure method of key node pairs. The general
idea is to delete the targeted node pair in the information networks instead of measuring it directly,
then indicators are designed to observe the changes of the whole networks. The degree of change
reflects the importance of the node pair, which results in a certain degree of equivalence in the
node pairs that have same networks variation.
Based on the deletion algorithm, the author proposes the LN and LS indexes respectively for
measuring the numbers of nodes (information stock) and the loss of link’s weight (information
flow) in order to find key node pairs in the undirected networks. An example of scientific
collaboration network related to the topic of h-index shows that the method has unearthed various
types of key node pairs that have similar structural equivalence, playing an organizational role or

* Correspondence should be addressed to ZHAO Xing, Email: xzhao@infor.ecnu.edu.cn, ORCID: 0000-0001-9347-590X