Blueprints for Text Analytics Using Python

Chapter 11. Performing Sentiment Analysis on Text Data

In every interaction that we have in the real world, our brain subconsciously registers feedback not just in the words said but also using facial expressions, body language, and other physical cues. However, as more of our communication becomes digital, it increasingly appears in the form of text, where we do not have the possibility of evaluating physical cues. Therefore, it’s important to understand the mood or sentiment felt by a person through the text they write in order to form a complete understanding of their message.

For example, a lot of customer support is now automated through the use of a software ticketing system or even an automated chatbot. As a result, the only way to understand how a customer is feeling is by understanding the sentiment from their responses. Therefore, if we are dealing with a particularly irate customer, it’s important to be extra careful with our responses to not annoy them further. Similarly, if we want to understand what customers think about a particular product or brand, we can analyze the sentiment from their posts, comments, or reviews about that brand in social media channels and understand how they feel about the brand.

Understanding sentiment from text is challenging because there are several aspects that need to be inferred that are not directly evident. A simple example is the following customer review for a laptop purchased on Amazon:

This laptop is full of series problem. ...