Natural Language Watermarking: Design, Analysis, and a Proof-of-Concept Implementation

摘要: We describe a scheme for watermarking natural language text by embedding small portions of the watermark bit string in syntactic structure number selected sentences text, with both selection and keyed (via quadratic residue) to large prime number. Meaning-preserving transformations (e.g., translation another language) cannot damage watermark. Meaning-modifying have probability, damaging watermark, proportional length over sentences. Having key is all that required reading The approach best suited longish meaning-rather than style-oriented "expository" texts reports, directives, manuals, etc.), which governments industry produce abundance need protection more frequently fiction or poetry, are not so tolerant meaning-preserving changes implements.