The rise of media and new tools in computer science provide new approaches to study culture. We proposed a novel research method that leverages facial expression and language features from TV series to assist cross-cultural studies. We first compared the statistical results of the features and drew a set of conclusions that can be supported by a number of previous works in cultural studies. Then, we employed the multimodal features to train a classifier to recognize the country that a TV series belongs to. A linear SVM achieved a high average accuracy of 94.4%. We further interpreted the coefficients of the model and obtained new observations from the results. Our method can avoid the drawbacks of the existing cross-cultural research approaches, provide new ideas for media data analysis, and aid culture adaptive product design.
CITATION STYLE
Xu, X., Qie, N., & Rau, P. L. P. (2021). Multimodal Features as a Novel Method for Cross-Cultural Studies. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 12771 LNCS, pp. 527–546). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-77074-7_40
Mendeley helps you to discover research relevant for your work.