Detecting emotions in text is an application of natural language processing (NLP) and an active area of research today. Whereas sentiment analysis is only one dimensional (positive/negative), emotion detection represents many as three dimensions (aka, the valence-arousal-dominance [VAD]  spectrum). People express emotions through their speech, facial expressions, gestures, and writings. There are several ways to computationally represent emotions: treat them as discrete categories, such as happiness or sadness, or using the VAD spectrum. 

Emotion detection can significantly increase the effectiveness of applications such as opinion mining or diagnosis of various mental disorders from human-generated content.

In this talk we will focus on emotion detection from text and cover some of existing emotion-detection techniques, available datasets, and the challenges of building a production-level emotion detection system.