The Study On Self-similarity Of Scientific And Technical Literature On Time Scale

With large-scale expansion of scientific and technical literature,literature data volume in a subject area becomes oversized that using the journals selecting or keywords searching to get data sample has become the most main way in the study of scientometrics.This approach is feasible because scientific and technical literature has self-similarity,that is the similarity of part and overall. However,the scientificalness of part reflecting overall has not been proved yet.As the important carrier of scientific knowledge,the scientific literature and patent documents has become the most main research objects and data sources of scientometrics.This paper set about from two aspects of scientific literature and patent documents,using various methods such as scientometrics,time series analysis,linear regression analysis and fractal geometry.The paper select four representative subjects and also select patent document which has more small calss and easy to quantitative analysis as cases to study.Compare the part and overall data.Through quantitative analysis of large amounts of data,time sequence diagram,scatter diagram and linear regression data table all show the sample data growth trend was consistent with the overall data, has high correlation,and fitting is very ideal.The self-similarity of scientific and technical literature on time scale can be fully validated by these results.The paper give theoretical evidence to data sampling of scientometrics.This paper sums up several classic growth model of scientific and technical literature and the deficiency of each model,then puts forward the understanding of self-growth of scientific and technical literature.It holds that the evolution of scientific and technical literature itself has self-similarity,namely between one phase and phase,and between one phase and the evolutionary growth process has self-similarity.The study on self-similarity on time scale in this paper reflects the self-similarity between part and the evolutionary growth process.Through the speculation and analysis of the results of case analysis,the paper summarizes the selection regularity of data sampling and qualitatively and quantitatively analyze relationship between the degree of self-similarity and scale.Through the introduction of the definition of fractal dimension,and selection of dimension calculation method,a validation method of the optimum scale of data sampling are obtained.Lastly,select the data of case analysis to verify the method.The paper provides reference and guidance for data selection.