Stylometric Watermarking

Using stochastic data to identify ownership has a long history, with stochastic features being used for a variety of identification applications such as voice print, handwriting recognition, and authorship attribution. Embedded traits such as stylistic preferences would help solve the question of ownership. Even though people sometimes make attempts to disguise their styles, they are rarely successful. Stylometry, a science of measuring literary style is often used in authorship attribution. A written piece is analyzed using statistics, math formulas and artificial intelligence to determine the "style" of an author\´s writing. The stylometric watermarking proposed in this paper tries to model the sty- lometry and apply it to digital watermarking. In the stylometric watermarking, the watermark is no longer deterministic, but stochastic, or a style.