Emotion recognition for Twitter language based on lingual and expressional information
Yoshie Setsu, Xin Kang, Shun Nishide, Fuji Ren · 2018
There have been many sentence-based and dictionary-based emotional corpora in the study of textual emotion recognition. However, many new expressive symbols, which we call the “facemarks”, have been observed in the social network language with the development of social network services (SNS). And studying the emotion expressions with these facemarks has been a new challenge yet a chance for emotion recognition in the social network. In this paper, we present our work on a study of emotion recognition for the Twitter language, by constructing a Twitter corpus with various kinds of facemarks in Tweets and evaluating the emotion expressions in these expressive symbols. By a parallel experiment of the emotion recognition based on the lingual and expressive features, we find that the expressive information has a decisive function in the recognition of Twitter language emotions.