机器学习应用于中国人口学研究:进展、框架与展望OACHSSCD

In the era of digital intelligence,the research methods in demography have been continuously innovated.Machine learning has expanded the traditional paradigm and research boundaries of population science,and improved the scientificity and accuracy of population governance,bringing about profound impacts on demography research.Based on bibliometrics and scientific knowledge mapping methods,this paper systematically reviews the evolution process,main contributions,common misunderstandings and development direction of machine learning in Chinese demographic studies.The research found that the applications of machine learning has gone through three stages:the early exploration phase,the stable accelerration phase,and the rapid explosion phase.Its focus covers the core research subjects of demography including birth,death and migration,as well as the major research topics in population studies such as population spatial distribution,aging and health,and human capital.By combining domestic and international research process,it constructs an analytical framework for applying machine learning to demographic research,covering the key modules of demographic data construction,phenomenon description,causal explanation,trend projection and policy evaluation.Currently,machine learning in demography still encounter such problems and limitations as sample selection bias,limited algorithmic interpretability,ethical risks and inadequate exploration of native contexts.Future efforts should be conducted in areas such as multi-source data integration,interpretable machine learning and algorithmic fairness governance,so as to promote the construction of a demographic research methodology system with Chinese characteristics,which is oriented toward addressing major issues of China's population development,driven by digital intelligent empowerment for method innovation,and integrated with theoretical foundations and practical values.