METHOD FOR IDENTIFYING IMPORTANT NODES IN THE CITATION NETWORK OF SCIENTIFIC PUBLICATIONS

Abstract

A mathematical method for selecting and ranking publications according to the degree of their compliance with the objectives of the research is developed and investigated. Bibliographic and abstract databases are used as primary sources of data on publications, which make it possible to track the citation of publications and identify the corresponding citation networks. The subject of the study is the citation networks of scientific publications. The mathematical model of citation networks is simple directed graphs whose vertices correspond to publications and whose arcs correspond to bibliographic references. The objectives of the research can be the preparation of a scientific article, writing a monograph or textbook, the design of a final qualifying work or dissertation, the replenishment of the library fund, etc. The research is carried out using the methods of Network Science. A method is proposed for determining the importance of the nodes of the citation network of scientific publications, taking into account the relevant measures of the centrality of the nodes and the profile of the research of publications. Relevant measures of centrality are indicators of the importance of nodes that are more appropriate than others and meet the search query for the selection of publications. The paper considers three profiles of the research of the citation network in order to rank publications. Since the citation network is oriented, the incoming and outgoing connections of the node are analyzed separately. The difference between the study profiles is that in one of them, only outgoing connections are taken into account in the centrality measures, in the other – incoming connections, and in the next, both are taken into account. An example of the application of the developed method of selection and ranking of scientific publications is given. The experimental values of the network node importance indicators were obtained on the basis of data on the citation of scientific publications in the RePEc bibliographic database.