Authors In The Field Of Information Science In My Country Co-authored Network Analysis

At present,we are in the age of big science,and the research activities are characterized by large investment intensity,multi-discipline crossing,expensive and complicated experimental equipment,and ambitious research objectives.Therefore,scientific research cooperation is the mainstream way of scientific research,and it is embodied in the co-author of scientific and technological paper.These co-authorships are intertwined and form a network of authors.The study of co-author network is helpful to find the co-authorship between authors,analyze the research topic of the research group,and explore the knowledge structure within the discipline.It can also provide suggestions for the co-authorship construction,predict the evolution trend of co-author networks,and promote better and faster development of scientific research activities.When research co-author network research status at home and abroad,this paper found that the current for study co-author network limited in measure the co-author network parameters,found that the study group,or pushed by time window to explore the evolution trend of the co-author network;There are also scholars working on the link prediction of the co-author network to provide suggestions for the establishment of the author's co-author relationship;However,few authors have combined the research topic and citation two factors to study the development trend of co-authored network.Therefore,this article puts forward the author's research topics into the co-author network based on citation(referred to as CAN-C),as a case study of information science disciplines,inquiry the structure and characteristics of the co-author network based on topic and citation(referred to as CAN-T&C).First of all,this paper selects 154 prolific authors in the field of information science in our country,CAN-C is constructed,analyzes the characteristic parameters and structure characteristics of the network,and it can be divided into seven modules,and analyzes the characteristics of the structure and content of each module,and the strength of co-authorship between modules.Secondly,use the LDA topic model,extracts the topic of 154 authors' document,generate the document topic of each author,then through the Pearson correlation analysis to generate the author similarity matrix based on topic(referred to as ASM-T).Then,the fusion function is constructed to integrate the ASM-T into the co-author matrix based on citation(referred to as CAM-C),then the CAN-T&C is constructed.Similar to the CAN-C,Based on the analysis of CAN-T&C related characteristic parameters and structural features,it is divided into 9 modules,further analysis of each module's internal structure and author link relationship,analysis of link strength between modules.Finally,the comparison between CAN-C and CAN-T&C is carried out from three aspects: network parameters,structural features and content characteristics.After the comparison of the two co-author network,it is found that compared with CAN-C,the density,clustering coefficient and average node degree of CAN-T&C have been improved in terms of network parameters;In terms of network structure,the number of modules of CAN-T&C has increased,and the correlation within and between modules has been strengthened;In terms of content,CAN-T&C is more accurate and comprehensive on reflects the author's topic relation.We believe that taking citation and topic into consideration when constructing co-author network will enable the co-author network to be enriched,improve the knowledge structure within the discipline,promote knowledge exchange within disciplines and the effective evolution of scientific research activities,and achieve better and faster development of disciplines.