Towards Text-based Emotion Detection A Survey and Possible Improvements

Abstract

This paper presents an overview of the emerging field of emotion detection from text and describes the current generation of detection methods that are usually divided into the following three main categories: keyword-based, learning-based, and hybrid recommendation approaches. Limitations of current detection methods are examined, and possible solutions are suggested to improve emotion detection capabilities in practical systems, which emphasize on human-computer interactions. These solutions include extracting keywords with semantic analysis, and ontology design with emotion theory of appraisal. Furthermore, a case-based reasoning architecture is proposed to combine these solutions.