The searchable internet contains almost 2 billion websites. And new, text-rich sites are being added at a rapid pace: more than 700 million popped up from 2016 to 2017, according to the International Real Time Statistics Project. A lot of this web-based text is relevant to marketers: online product reviews, information about purchasing behavior, customer-to-customer interactions, and transcribed tele-sales calls. Marketers now have more information from consumers in the form of written words than ever before. The problem, as with any extremely large data set, is determining how best to use the information. The relatively new fields of text analytics and sentiment analysis offer marketers a solution, enabling them to turn vast amounts of emotion-rich, word-based data into actionable information about consumers.
This note explores dictionary-based sentiment analysis using programming language R; it also introduces empirical sentiment analysis.