Sentiment analysis is merely another term given to text classification or document classification – where the classifying feature happens to be the sentiment of the text. We can understand sentiment as a feeling or opinion about something – if we said The movie was terrific!, it means it expresses a positive sentiment or feeling, and if we say The movie is terrible!, it would be expressing negative sentiment or feeling. Here, sentiment usually refers to positive or negative sentiment, but this can, of course, be extended to include multiple sentiments, such as angry, sad, happy, and maybe even a thoughtful sentiment if we so wish. In other words, sentiment analysis tasks are simply classification tasks where each class is a kind of sentiment which we wish to analyze.
In fact, we have seen an example of sentiment analysis in the previous chapter, when we used Keras and spaCy together to build a deep learning pipeline. Sentiment analysis was performed by assigning probability...