科学学的理想、演进与走向

The Ideal,Evolution and Trend of"Science of Science"

科学学自诞生以来,一直致力于对整体的科学开展精确的、定量的研究,以形成对科学的客观认识。几十年来,科学学研究一直遵循着计量科学学的范式,经历了基于数理统计、基于引文网络再到基于知识图谱的科学计量研究。目前,随着计算机技术发展和计算社会科学的兴起,科学世界的量化表征数据类型逐渐丰富多元,特别是计算能力提升和计算逻辑发生变革,科学学在“计量科学学”研究范式基础上拓展到“计算科学学”的新范式、新方向,为科学学理想的实现提供了新的可能。

Since the birth of Science of Science,it has been committed to carrying out precise and quantitative research on the overall science in order to form an objective understanding of science.Since the 21st century,information technologies such as big data have been increasingly embedded in the process of scientific research.The digitization,complication and networking of scientific activities continue to deepen.The data types are gradually diversified and the scale is massive,etc.All these have brought new possibilities for the development of Science of Science.The ideal of Science of Science is to use scientific methods to form a reliable and objective understanding of the scientific world,and to provide a reliable basis for scientific and technological decision-making.To benchmark the ideals of Science of Science,from the perspectives of quantification and computation,focus on three issues to analyze(1)the evolution of quantitative representations of the scientific world.(2)the evolution of measurement logic for various types of quantitative data,(3)possibilities offered by the diversification of quantitative data and improvement of computing power.Summarize the development characteristics and deficiencies of Science of Science from the above three perspectives.From the beginning of the 20th century to the present,Science of Science has always followed the paradigm of"Metrological Science of Science",using precise and quantitative methods to study,explore,and recognize the scientific world.The three stages are mathematical statistics,citation networks,knowledge mapping.In recent years,with the rise of computational social science.The representation scope and method of the scientific world and the calculation method of representation data have all changed,the Science of Science has entered the stage of"Computational Science of Science"."Computational Science of Science"is a new research paradigm of Science of Science that realizes a panoramic cognition and systematic understanding of the scientific world through mining and analysis of large-scale data sets about scientific and technological innovation activities."Computational Science of Science"has two characteristics.First,data types are no longer limited to structured data such as bibliographic information such as papers or patents,index data,etc.,but are extended to large-scale structured data or unstructured data that reflect different aspects of scientific activities.The second is the change of calculation method and calculation logic.Algorithm-driven is the most distinctive feature,while the algorithm is non-standardized and needs to be designed and constructed according to research problems and data characteristics.At present,some studies have been carried out by means of"Computational Science of Science"and have reached conclusions that are different from traditional understandings.Using scientific methods to study science is the ideal of Science of Science,which is also the internal driving force for the development of Science of Science,and promote the transformation of the research logic of Science of Science from"Metrological Science of Science"to"Computational Science of Science".The difference between the two is that the former is limited to data types and computing capabilities,and it is a relatively closed research that requires the help of scientific indicators provided by databases,existing software and standardized algorithms;The latter is an active and open research paradigm based on a large amount of overall data,active algorithm design,and realization of relationship analysis between variables,causal relationship inference,etc.Although there are still some shortcomings in"Computational Science of Science",for example,the systematicness of quantitative representation of the scientific world needs to be improved,and the pertinence,stability,scalability,and limitations of algorithms must be continuously improved.Overall,The development of"Computational Science of Science"has made a big step towards the ideal of science of science,which is closer to the ideal at the beginning of Science of Science than"Metrological Science of Science".Facing the future,Science of Science should focus on the combination of various research approaches,including"Computational Science of Science",in order to gradually achieve a"holistic understanding"of science by combining the understanding of science formation from different perspectives.