社会科学研究中的“等价共同体”问题

The Problem of“Equivalent Communities”in Social Science Research

“等价共同体”是社会科学研究的重大关切问题。在长期的社会科学研究实践中,人们通常会忽略一个深层次的“等价共同体”问题。要么认为不存在“等价”问题,要么认为“等价”是理所当然的。“等价共同体”综合考察社会科学研究的样本代表性、分布特征、行为机制、干预反馈、结论稳健性等多维度品质,应予以系统性梳理与研究。在各阶段,“等价共同体”理念存在不同的实践形式。在社会调查研究阶段,“等价共同体”形成雏形,表现为样本的集合,包括普查、抽样调查、互联网调查等。在计算社会科学阶段,“等价共同体”深入发展,表现为数据共同体,在样本数量、数据维度方面大为拓展,应用于互联网文本挖掘、自然语言处理、在线社交网络、机器学习、地理信息系统等领域。在智能体模拟时代,“等价共同体”问题无法回避,体现为智能体集合,并继承了社会调查(定量实证研究)、计算社会科学(大数据)基础。通过不断训练、锤炼,最终获得和真实社会在多个维度(调查数据、大数据等)高度匹配的“等价智能体”,开展社会分析与社会治理。一个总体性趋势是,“等价共同体”属性的被满足程度在逐渐增强。

“Equivalent community”is a major concern and core requirement in social science research,and should be systematically studied.In long-term social science research practice,people often overlook a deepseated issue of“equivalent community”.Either they think that there is no“equivalence”problem,or they think that“equivalence”is a matter of course.“Equivalent community”comprehensively examines the multi-dimensional qualities of social science research,such as sample representativeness,distribution characteristics,behavioral mechanisms,intervention feedback,and conclusion robustness.“Equivalent community”has different practices in various stages of social science research.During the social survey stage,the“equivalent community”takes shape,which is represented by a collection of samples,including censuses,sample surveys,Internet surveys,etc.During the stage of computational social science,the“equivalent community”has developed in depth and manifested as a data community,which has been greatly expanded in terms of sample size and data dimensions,and has been applied to Internet text mining,natural language processing,online social networks,machine learning,and geographic information systems and other domains.During the stage of agent simulation,the issue of equivalence cannot be avoided.The“equivalent community”is embodied as a collection of agents and inherits the foundation of social survey(quantitative empirical research)and computational social science(big data).Through continuous training and refinement,we will eventually obtain“equivalent agents”that is highly consistent with the real society in multiple dimensions(survey data,big data,etc.)to analysis and governance.A general trend is that the degree to which the“equivalent community”nature is satisfied is gradually increasing.