The Study Of Scientific Community From Viewpoint Of Complex Network

With steady economic development of our country and constant increase in fund investment from the government,more and more people are engaged in scientific research.Stable communities of researchers taking scientific research as a career have emerged in various disciplines.Formed by researchers from the same or relevant research field,scientific community is not only the producer of science and knowledge,but also the creator of value and ethic of science,playing a crucial role in the development and progress of science.During the development of modern science,scientific community takes much diversified forms in society.Effectively identifying scientific community and studying on its structure,correlation and evolution is fundamental to effective management and scientific improvement of scientific community.Taking all domestic studies on scientific community into consideration,most of them are limited to a theory level and lack empirical study revealing internal structure and activity patterns.As a socialized system with specified functions,scientific community has basic characteristics of a complex system.Complex network is an important branch of systematic theory research,and also a common method of studying complex systems.The analysis theory of complex network offers a new point of view for study of science community.Firstly this dissertation introduces background and significance of the study.Afterwards,it defines the object of study and elaborates overview of both domestic and foreign study of scientific communities and scientific collaboration networks.Through the overview,three defects are discovered in current study:(1)Lack of unified theory and method of community classification.(2)Insufficient attention to details study of correlation between scholars,instances including Authors' signatures are seldom computed with signing order,in addition only collaborative relation is studied,while other relations are ignored.(3)Though lots of study regarding scientific research collaboration network can be seen in our country recently,there are very few approaches to collaborative relation from aspect of chronological evolution.Compared with study abroad,there is great disparity in terms of evidence quantity and processing method.Starting from the point of to overcome the deficiency mentioned above and approaching from point view of complex network,this dissertation treats scientific community as a complex social system and analyzes the structure of scientific community and characteristics of its dynamic evolution.The dissertation attempts to derive behavior feature and activity pattern of community members from study of network structure evolution,and construct a complete framework for analysis of dynamic evolution of scientific community,aiming at offering a new theory and method of study.Study in this dissertation includes the following sections:?.Identification of scientific community based on cluster analysis(1)Considering basic characteristics of literature reference data,the dissertation designs a framework model of repeated-name-elimination based on pattern and similarity,along with elimination algorithm.Experiments on data of three principals all show the effective identification rate achieved through this elimination algorithm as more than 90%.Moreover,the bigger the data is,the better result can be achieved.(2)By fully utilizing analysis parameters of cluster processing tool and combining them with characteristics of data,the dissertation establishes evaluation criteria for cluster algorithm output,which introduces a new theory for clusters number decision and a concrete method of scientific community identification.?.Construction of complex network for scientific community and analysis of its characteristics(1)The dissertation sets up an algorithm model for scholars' correlation,which combines scholarly collaboration activities and scholarly citation activities and describes scholars'correlation from multi aspects by interlacing the relation of literature coauthorship and the relation of literature citation.(2)The dissertation elaborates coauthorship details by discriminating the lead author from other coauthors,thus transforms conventional collaboration network undigraphs into digraphs.A complex network of scientific community constructed in this way contains more information,which is helpful to analysis in different levels and from different angles.(3)From viewpoint of complex network,the dissertation compares the overall network constructed and various scientific community networks derived from cluster analysis,including internal structure and external relationships two aspects?.Study of dynamic evolution of scientific community on basis of complex network(1)The framework of dynamic evolution analysis is constructed through integrating study in above sections.(2)The dissertation puts forward a new method of time slicing in a sliding time window.Empirical analysis of characteristics of dynamic evolution of scientific community network structure in different time slices is performed on basis of data of published literature in 12 years(3)Network nodes are categorized into 3 types.A formula is defined based on the idea of relative entropy to measure the network variation.Empirical study of quantitative change of scientific community network during evolution is made(4)The dissertation discusses correlation between article amount of authors and in-degree of nodes during network evolutionThis dissertation features the following innovation(1)The dissertation introduces technology of cluster analysis and complex network into study of scientific community.It constructs an analysis framework for dynamic evolution of scientific community,and performs empirical study on basis of literature data from library and information science.And it offers a new feasible method of studying scientific community.(2)Based on the theory of behavior-structure-function model proposed by itself,this dissertation makes various novel attempts and exploration,such as analysis of network structure and it's evolution,analysis of internal structure and external relationships for each divided science community,analysis of stability of in-degree/out-degree of nodes and network itself duringnetwork structure evolution using time sequence method,analysis of stability of correlation,and analysis of position shift among collaboration,proposition of a new studying method of depicting group behavior in scientific community.