ThoughtMap: Visualizing Cognitive and Philosophical Structures Through Large-Scale Lyric Embedding and Semantic Clustering

Language is often assumed to reflect personality, values, and cognitive tendencies. However, few tools exist that allow individuals to visualize and compare their own patterns of thought at scale. This study introduces ThoughtMap, an experimental framework for extracting, clustering, and visualizing semantic structures from large collections of written works. A dataset consisting of 992 self-authored song lyrics was embedded into a shared semantic space using transformer-based language embeddings. Similarity search, unsupervised clustering, and dimensionality reduction techniques were applied to identify recurring thematic and cognitive structures. The resulting visualization revealed several major domains, including Society, Cognition, Experimentation, Causality, Introspection, Philosophy, and Technology. The findings suggest that large collections of creative writing may contain measurable patterns that extend beyond conventional genre classification and may instead reflect persistent cognitive and philosophical tendencies. The long-term objective of the project is to develop a searchable framework for comparing thought structures across individuals and identifying similarities in cognitive profiles.