With Emotion Analysis, you can bring sensitivity to analytics. Stay attuned to the feelings of customers during chat sessions, track social media reactions to your press releases, or gauge the public outlook on financial news. Emotion Analysis can infer emotions from text samples of any size up to 50KB.
Input posted text, posted HTML, or a link to a web page, and Emotion Analysis will leverage sophisticated machine learning to return confidence scores for anger, disgust, fear, joy, and sadness. Each score indicates the probability that the corresponding emotion is implied by the sample text.
Emotion Analysis uses a stacked generalization based ensemble framework to derive the final emotion scores from initial results of several machine learning models. Each lower level model uses a unique combination of machine learning algorithms and language features such as words, phrases, punctuation, and overall sentiments.
Note: Emotion Analysis is currently in beta. The Emotion Analysis Beta will end on Friday, July 1st 2016. On that date all users will start to be charged per event for the Emotion Analysis API.
Emotion results for a web page
Emotion Analysis currently supports English content.
Emotion data can be returned in either JSON or XML to fit the needs of your application.