A Systematic Review on Responsible Multimodal Sentiment Analysis in Marketing Applications

The intrinsic challenges of contemporary marketing encourage discovering new approaches to engage and retain customers effectively. As the main channels of interactions between customers and brands pivot between the physical and the digital world, analyzing the outcome behavioral patterns must be achieved dynamically with the stimulus performed in both poles. This systematic review investigates the collaborative impact of adopting multidisciplinary fields of Affective Computing to evaluate current marketing strategies, upholding the process of using multimodal information from consumers to perform and integrate Sentiment Analysis tasks. The adjusted representation of modalities such as textual, visual, audio, or even psychological indicators enables prospecting a more precise assessment of the advantages and disadvantages of the proposed technique, glimpsing future applications of Multimodal Artificial Intelligence in Marketing. Embracing the Preferred Reporting Items for Systematic Reviews and Meta-Analysis as the research method to be applied, this article warrants a rigorous and sequential identification and interpretation of the synergies between the latest studies about affective computing and marketing. Furthermore, the robustness of the procedure is deepened in knowledge-gathering concerning the current state of Affective Computing in the Marketing area, their technical practices, ethical and legal considerations, and the potential upcoming applications, anticipating insights for the ongoing work of marketers and researchers.