Controllable Semantic Linguistic Steganography via Summarization Generation

In recent years, linguistic generative steganography has been greatly developed. Current linguistic steganography methods can generate sufficiently fluent steganographic text, but cannot control the semantic expression of the generated steganographic text. In social network, steganographic text with inconsistent contextual semantics may pose the vigilance of the regulators. This paper proposes a controllable semantic linguistic steganography method via summarization generation. By analyzing the semantic content of the given text, a compact summarization that retains the most important content is generated, which can ensure the consistent semantics. During summarization generation, the secret information is embedded based on the dynamic programming algorithm to find optimal steganographic encoding path. Compared with the existing methods, the experimental results show that the proposed method has greater advantages in semantic consistency while ensuring security.