In this paper we address the problem by using several compression-inspired strategiesthat generate different models without analyzing or extracting specific features from thetextual content, making them style-oblivious ...