Using Markov chains for identification of writers
D. Khmelev · Literary and Linguistic Computing · 2001
In this paper we present a technique for authorship attribution based on a simple Markov chain of letter, i.e. only letter bigrams are used. Many proposed methods of authorship attribution are illustrated on small examples. We show that this technique provides excellent results when applied to over 380 texts from the Project Gutenberg archives, as well as to two previously published datasets.