Authors: Dr. Ravi I A, Dr. Saley Seetharaman
Abstract: The intersection of neuromarketing with social media analytics brings new possibilities for understanding the hidden motivations of consumers and making predictions about their purchasing behavior. This paper introduces a unique approach to creating an integrated model that incorporates information from both EEG readings and social media sentiments. Based on the multimodal data from 250 participants (EEG, eye tracking, GSR) and 5 million messages collected from social networks regarding 12 different products, this study reveals cognitive and affective biometric indicators that reflect consumers' preferences. Then, a Hierarchical Attention-based Deep Neural Network (HA-DNN) is developed to incorporate neural biomarkers of preferences and sentiments expressed on social media and predict individual purchasing decisions. This model shows 86.7% accuracy in predicting purchase decision with AUC = 0.91 and outperforms social media sentiment models (71.2%) and neural network models (78.4%) substantially.
DOI: https://doi.org/10.5281/zenodo.20416237
