Linguistic Steganography and Linguistic Steganalysis

The objective of linguistic steganography is to embed additional data in text carriers for covert communication, whereas linguistic steganalysis, as a counter technology to linguistic steganography, aims at revealing the existence of additional data within unknown texts. Early linguistic steganography algorithms alter a text carrier to embed additional data, which limits the embedding payload due to the extremely small number of changeable operations. With the rapid development of deep learning and natural language processing techniques, recent advances in linguistic steganography utilize a well-trained language model to directly produce the text carrying additional data without a given text, which results in a significantly higher payload when compared with previous works and enhances the level of security. On the other hand, early linguistic steganalysis algorithms rely heavily on manually crafted statistical features, which cannot well reveal steganographic characteristics and thus limit the detection accuracy. Recent advances demonstrate that by modeling the local and global semantic, syntactic and contextual characteristics through deep learning, discriminative features can be extracted for high-accuracy detection. To trace the latest developments and trends, in this chapter, we will review advanced methodologies in linguistic steganography and linguistic steganalysis, and discuss challenges and opportunities. This chapter is intended as a tutorial of text based steganography and steganalysis in the context of media security.