Textual Emotion Processing From Event Analysis

Textual emotion recognition has gained a lot of attention recent years; it is however less developed due to the complexity nature of emotion. In this paper, we start with the discussion of a number of fundamental yet unresolved issues concerning emotion, which includes its definition, representation and technology. We then propose an alternative solution for emotion recognition taking into account of emotion causes. Two pilot experiments are done to justify our proposal. The first experiment explores the impact of emotion recognition. It shows that the context contains rich and crucial information that effectively help emotion recognition. The other experiment examines emotion cause events in the context. We find that most emotions are expressed with the presence of causes. The experiments prove that emotion cause serves as an important cue for emotion recognition. We suggest that the combination of both emotion study and event analysis would be a fruitful direction for deep emotion processing.