Water marking in Turkish: A Syntax Based Model

The aim of this study is to develop a natural language watermarking algorithm based on morphosyntactic manipulations of a text. This watermarking modality was chosen based on the observation that the Turkish syntax allows for a more stable and rich repertoire of tools. The watermarking algorithm utilizes linguistic tools such as dependency parser, tree converter, Wordnet and a lexicon. It embeds message bits in the sentences based on the applicability of tools per sentence, their choice being randomized for stylistic and security concerns. We give the statistics of the tool occurrences and measure their acceptability based on edit counts of readers.