Esteganografía lingüística en lengua española basada en modelo N-gram y ley de Zipf

Linguistic Steganography is a science that utilises computational linguistics to design systems that can be used to protect and ensure the privacy of digital communications and for the digital marking of texts. Various proposed ways of achieving this goal have been documented in recent years. This paper analyses the possibility of generating natural language texts in Spanish that conceal information automatically. A number of hypotheses are put forward and tested using an algorithm. Experimental evidence suggests that it is feasible to use N-gram models and specific features of the Zipf law to generate stegotexts with a good linguistic quality where human readers could not differentiate the stegotext from authentic texts. The stegotexts obtained allow the concealment of at least 0.5 bits per word generated.