Segarra, Eisen, and Ribeiro present a forensic‑linguistic authorship attribution technique that models the relational structure of function words as normalized adjacency networks interpreted as Markov chains, comparing texts via relative entropy to capture stylistic signatures distinct from frequency‑based methods; numerical experiments on corpora of up to 18 authors show attribution accuracies of ≥90 % and demonstrate complementary gains when combining relational and frequency information.
Bears on: authorship & stylometry · concepts & methods