This paper investigates forensic e‑mail authorship verification by adapting the NIST speaker‑recognition evaluation framework to textual stylometric features, evaluating the approach on a real‑world e‑mail corpus and reporting average equal‑error rates of 15–20 % and a minimum detection‑cost function of 0.0671. The work contributes to authorship verification and stylometry‑based attribution methods within e‑mail forensics.
Bears on: authorship & stylometry · concepts & methods