An Introduction to the Possibility of Applying the k-means Algorithm to the Typology of Characters from Folk Tales

Abstract

The classification of folktale characters through the application of statistical methods – particularly those associated with unsupervised machine learning – remains a largely unexplored domain within folklore studies. Several intellectual traditions may be regarded as providing the conceptual foundation for such an inquiry, including formalism, structuralism, and cognitive approaches to the study of mythical forms, all of which share an emphasis on identifying recurrent patterns in folklore texts. In recent years, clustering algorithms have undergone significant development. Their current state of advancement appears sufficient to support the basic requirements of introductory analyses in folklore studies. The study presented in this article was designed to yield unequivocal and interpretable results, and this objective was successfully achieved. While the findings suggest that the application of clustering algorithms to folklore research holds considerable promise, the issue of continuity between the resulting groups remains problematic. Addressing this challenge calls for further theoretical reflection and the development of methodological frameworks capable of accommodating the fluid and overlapping nature of folktale character classifications.