Sentimient Analysis extracts the sentiment polarity, subjectivity, irony and emotional agreement expressed in a text. The global polarity can take one of the following six values: strong positive, positive, neutral, negative, strong negative or none, for the cases where no sentiment is expressed. The analysis is done using the General domain sentiment model provided in the Sentiment Analysis API.
These are the settings for this recipe:
The output dataset will have five new columns: "sentiment_polarity", "sentiment_subjectivity", "sentiment_irony", "sentiment_agreement" and "sentiment_confidence". The following example uses the dataset used in this tutorial. The texts are in English.