学习情感计算研究——基于国际研究的系统性文献综述

Learning Affective Computing Research——A Systematic Literature Review Based on International Studies

情感计算为感知与理解学习者情感、增强情感交互、促进人机协同提供了技术支撑。近年来,围绕学习情感计算的研究迅速增加,但缺乏对相关研究理论基础、技术方法和应用场景的系统梳理与总结。鉴于此,本研究采用系统性文献综述法对国际上发表的该主题研究进行内容分析,旨在为我国学者开展相关研究提供更多的实践参考。分析发现,学习情感计算研究以离散型情感理论为主,基本情感理论是研究焦点;情感测量的数据来源丰富,多模态情感识别成为研究趋势;情感测量方法以机器学习技术为主,支持向量机和卷积神经网络是应用最多的两种算法;学习情感计算的应用场景单一,以在线学习场景为主。今后我国学者在开展相关研究时,需要重视情感理论的情景化、数据的多源化、方法的多样化、应用场景的丰富化,拓展学习情感计算研究的深度和广度。

Affective computing provides technical support for sensing and understanding learners’emotions,enhancing emotional interaction,and promoting human-computer collaboration.Research related to learning affective computing has increased rapidly in recent years,but there is a lack of systematic compilation and summarization of the theoretical foundations,technical approaches,and application scenarios.In view of this,this study adopts a systematic literature review method to analyze the content of the internationally published studies on the related topic,aiming to provide more practical references for scholars in China to carry out related research.It is found that the research on learning aff ective computing is dominated by discrete aff ective theory,and the basic aff ective theory is the focus;the data sources of aff ective measurement are rich,and multimodal affective recognition has become a research trend;the affective measurement methods are dominated by machine learning techniques,and support vector machine and convolutional neural network are the two most applied algorithms;the application scenarios of learning affective computing are single,which are dominated by online learning ones.In the future,when scholars in China carry out related research,they need to pay attention to the contextualization of emotion theory,the multi-source of data,diversifi cation of methods,and enrichment of application scenarios,so as to promote the depth and breadth of the research on learning aff ective computing.