Author: Adriashkin Roman
Date: 2026 (v0.1)
Abstract: We introduce a mathematical framework for representing text samples by compact finite-dimensional profiles, termed cognitive fingerprints. The framework is based on a feature map from token sequences into a Euclidean space of empirical statistics. This representation supports quantitative comparison of text samples through profile distances and similarity measures, as well as perturbation-based questions about stability under bounded edits.
Links: PDF | LaTeX source | arXiv submission pending