针对情感计算领域的情感描述和测量问题,系统分析了情感的范畴观和维度观在情感计算领域中的意义。采用传统的心理测量方法,通过分析人们对330张图片在16个情绪维度上的评分,综合评价已有心理学研究中所涉及的众多维度,构建了2维和3维情感空间,指出了3个维度的本质、对人类情感信息的表达精度和相应的命名方法。实验结果表明,情绪范畴在3个情感空间中可以很好地分离开来,而且它们在情感空间中的分布可以非常直观地展示出各个基本情绪范畴之间的关系。这一结果证明3维情感空间可以充分地表达和量化人类情感,是情感计算研究的基础。
The study addressed the emotion measurement in affective computing area, and construed the significance of the discrete emotional categories approach and the dimensional approach. The traditional psychological measurement method was used to collect the rating scores of 330 pictures on 16 emotional dimensions proposed by the previous psychological research. The results showed PAD (Pleasure-Arousal-Dominance)dimensions were sufficient and essential to represent the emotional states. The study further proposed the dimensional approach is based on the quantitative analysis, and it is the main approach to solve the computability of emotions. The model of PAD dimensions is essential to promote the affective computing research.