The fast growth of social robotics (SR) has not been unidirectional, but rather towards a multidisciplinary scenario, creating a need for collaboration between different fields. This divergent expansion calls for a clear analysis of the field aimed at better orienting the research, thus paving the future of social robotics. This paper aims at understanding how the SR research field evolved in the last two decades by analyzing academic publications in SR and human–robot interaction using natural language processing (NLP) techniques. The analysis spotted an overlap between SR and human–robot interaction research fields that have been disambiguated using a data-driven approach that leads to the identification of a new group of papers we clustered under the concept of “soft HRI.” This research topic has been analyzed by extracting trends and insights. Finally, another topic modelling step has been applied to identify seven sub-topics that have been discussed and analyzed picturing the current state of the art of SR. The paper reports a complete overview of the SR research field identifying various topics and sub-topics helping researchers in understanding the evolution of this field, thus supporting the strategic placing and evolution of their research activities.

社交机器人（SR）的快速增长并非单向的，而是朝着多学科的方向发展，这就需要在不同领域之间进行协作。这种不同的扩展要求对领域进行清晰的分析，以更好地确定研究方向，从而为社交机器人技术的发展铺平道路。本文旨在通过分析自然语言处理（NLP）技术在SR和人机交互方面的学术出版物，来了解SR研究领域在过去二十年中的发展情况。分析发现SR和人机交互研究领域之间存在重叠，这些领域已经使用数据驱动的方法消除了歧义，从而确定了我们在“软HRI”概念下聚集的一组新论文。通过提取趋势和见解对本研究主题进行了分析。最后，另一个主题建模步骤已应用于确定七个已讨论和分析的子主题，这些子主题描绘了SR的当前状态。本文报告了SR研究领域的完整概述，确定了各种主题和子主题，从而帮助研究人员了解该领域的发展，从而支持其研究活动的战略定位和发展。