Off-the-Shelf Technologies for Sentiment Analysis of Social Media Data: Two Empirical Studies
Arthur Carvalho, Lucas Harris · Journal of the Association for Information Systems · 2020
Off-the-shelf technologies provided by major cloud platforms promise to facilitate and democratize the use of artificial intelligence techniques. Organizations can now apply highly sophisticated, pre-trained models in a variety of situations, such as when analyzing the sentiment behind social media posts. Among other uses, this enables organizations to better understand their consumers’ opinions regarding products and/or services. In this paper, we first review technologies for sentiment analysis provided by major cloud platforms. We then compare the accuracy of these technologies against a technique widely used in managerial and information systems studies, namely the bag-of-words approach. Our two empirical studies use social media data collected from Twitter (short posts) and Facebook (long posts). We find that all the studied off-the-shelf technologies for sentiment analysis are vastly more accurate than the bag-of-words approach. We conclude the paper by discussing our results in light of the recent rise of low/no-code development practices.