The Chinese VAI (CVAI) corpus is a three-dimensional corpus contains 1465 sentences that have positive expressions but negative sentiment, wherein 1005 of the sentences were collected from the NTU Irony Corpus and the other 460 were collected from social media.
Each sentence is annotated with a real value from 1 to 9 on the valence, arousal and irony dimensions. The valence dimension represents the degree of positive and negative sentiment, arousal represents the degree of calm and excitement, and irony represents the degree of irony. All dimensions range from 1 to 9. Values of 1 on the valence, arousal, and irony dimensions, respectively, denoted extremely high-negative, low-arousal, and low-irony sentiment. Alternatively, values of 9 denoted extremely high-positive, high-arousal, and high-irony sentiment, and values of 5 denoted a neutral, medium-arousal, and medium-irony sentiment.
Please cite the following papers if you use the Chinese VAI corpus:
Click here to download the CVAI corpus.
Liang-Chih Yu
Professor
Department of Information Management, Yuan Ze University
lcyu@saturn.yzu.edu.tw