3DText: Perceiving sentence-level text on 3-D model of emotions

Abstract

Emotion is a psychological process which reveals the sentiments and feelings of a human being. Relating emotions detection process with psychological theory of emotions serves as the strong foundation for the system. In this paper, a model, 3DText, is proposed foe textual emotion detection. VAD (Pleasure, Arousal, and Dominance), a 3-D theory of emotion is used to extract features (P-A-D) from text. For this purpose, the dataset ANEW and WordNet are used. The objective of this paper is to determine VAD values at the sentence level of any size using word level VAD values for domain independent text. The proposed approach is evaluated on ISEAR and EMOBANK datasets. To the best of authors’ knowledge, no such model exists till date.

Keywords

• wordnet
• emotion
• vad values
• anew dataset
• sentiment
• dimensional theory of emotions
• rmse
• sme