Emotion detection state of the art

Abstract

Emotion recognition and analysis has been extensively researched in neuroscience, psychology, cognitive science and computer science. In the absence of verbal and facial cues commonly associated with recognizing emotions, can emotions be detected via intelligently analyzing textual information? Our main goal is to answer this question by presenting a state-of the- art overview of studies that have been carried out in the domain of text based emotion detection. To this end, we will explore how emotions are detected from textual sources. We begin with a discussion of currently prevailing emotion theories that provide the basis for text based emotion detection. To achieve an in depth and scientifically structured evaluation, we study the underlying structure of several approaches in the literature and design an evaluation framework. Existing approaches are critically evaluated against the proposed evaluation framework. In particular, the results of various systems are discussed and contrasted. The outcome of such a detailed analysis highlights the current gaps in scientific literature in the field of automated emotion detection.