| Abstract | Previous research has found that sentiment analysis is the focus of Artificial Intelligence (AI). Nevertheless, a limited number of studies exist on advanced AI applications. First, this study contributes to this limited literature. It is argued that Transformer Architecture has initiated a new phase in AI and sentiment analysis. We contribute to this argument by providing evidence that transformer-based AI provides better performance and capabilities. We do so by systematically reviewing and analyzing the applications of Transformer-based models for sentiment analysis. The focus of this paper lies in the broader marketing field. In addition, marketing is one of the most essential fields in business. The choice of marketing in examining Transformer-based sentiment analysis constitutes a novelty. Using the PRISMA methodology we reviewed 481 papers. Only eight papers fulfill the criteria. It is a somewhat surprising finding that so few papers have applications of Transformer-based AI for sentiment analysis in marketing. Most papers follow older AI methods (i.e., Recurrent Neural Networks, Convolutional Neural Networks, etc.). This leaves a gap in this area, and we examine novel trends and directions for future research. |
|---|