【作者简介】吕鹏，中南大学社会计算研究中心、北京大学武汉人工智能研究院

【文章来源】《社会学研究》2024年第4期

【内容提要】社会复杂系统的运行是一个多层次的“行动—结构”二元互构过程。当前理论研究提供了行动与结构二者关系的主观诠释，却往往得不到客观过程的证据支撑；定量研究能够客观揭示状态、趋势方面的变化，但难以收集行动与结构联动的直接数据。智能体仿真模拟（ABM）能够综合理论阐释的灵活性与定量研究的客观性，通过机制研究行动，通过涌现研究“从行动到结构”，通过学习研究“从结构到行动”，推动二元互构研究。ABM通过真实结果拟合和反事实推演拓展了社会学的知识边界。ABM应与理论研究、定量研究联动，共建计算社会学方法生态，助力学术共同体的形成。

【关键词】计算社会学；多智能体仿真；ABM仿真模拟；行动—结构互构

【项目基金】本文系中央引导地方发展专项“‘全息东湖’智能社会模拟器研发”（2023EGA035）的阶段性成果。

【全文链接】https://shxyj.ajcass.com/Magazine/show/?id=96629

Agent-Based Modeling: Advancing Efforts in Action-Structure Mutual Constructionl Research

Abstract: The running of complex social system takes on the mutual construction of action and structure, which is a dual process with multiple levels. Current theoretical research provides subjective interpretations of the relationship between the two, but often cannot be supported by evidence of objective processes; quantitative research can objectively reveal changes of states and trends, but it is difficult to collect data on the linkage of action and structure. Agent-Based Modeling (ABM) can integrate the flexibility of theoretical research and the objectivity of quantitative research to advance the mutual construction research. It uses the mechanism design to simulate individual actions, uses the emergence concept to study the process "from action to structure", and uses the learning tool to study the process "from structure to action". ABM largely expands the knowledge boundaries of sociological studies through fitting real-world outcomes and inferencing counterfactual outcomes. ABM should combine theoretical research and quantitative methods to build a methodological ecology of computational sociology and to help form and develop a domestic academic community.