Uncovering Hidden Narratives: Discovering and Classifying Archetypes in Tiny Stories

This study explores classifying narrative archetypes in a dataset of tiny-stories with the aid of natural language processing and machine learning techniques. Among the narrative archetypes are, The Adventurer, The Animal Companion, The Caregiver, The Curious Learner, The Innocent Child, The Loyal Friend, and The Trickster. These archetypes were derived using thematic keywords by the TF-IDF method. It follows a well-structured workflow: data cleaning and collection; exploratory data analysis (EDA); key phrase extraction, followed by labeling; vectorization of the data points; model building; and the evaluation. Such stories that fall outside these preset archetypes become outliers. Thus, the interpretation of the thematic flavor of each one of these will help in better understanding the extent and spread of narrative archetypes. This research has potential applications in literary analysis, content recommendation systems, and creative storytelling frameworks.