The Research Of Natual Language Watermarking

The development of computer network and multimedia technology provides opportunity to the distribution and delivery of digit products, but also bring some challenges. Nowadays, the protection of copyright of multimedia digit products is attracting high attention. Digit watermarking is widely believed to be a valid means to protect the copyright from this phenomenon.At present, there are a lot of document watermarking which focus on formats of document. These kinds of document watermarking are not very robust. Natural language watermarking can embed watermark into text content, it has better robustness, but the capacity is rather slow. This paper proposed an improved syntactic watermarking and an improved semantic watermariking to improve the capacity of watermark and assure the quality of text.The main work of this paper is as follows:The first part presents an improved method of syntactic translation of the original syntactic watermarking. The original syntactic watermarking uses constituent tree to analyse syntactic structure. Constituent tree produced by link parser is rather coarseness, so it will result in some shortages. This paper uses the more precise linkage tree and rules of words to change the syntactic structure of a sentence. The improved method can improve the capacity of watermark and can avoid some improper translation.Next, the paper analyses the syntax and the linkage tree of a sentence, then add some new translation to change syntactic structure. In a way, these new translation can help to improve the capacity of watermark.Then, the paper introduces syntactic translation into semantic watermarking according to the characteristic that modifying the syntactic structure will not change the semantic structure of a sentence, and proposed an improved semantic watermarking.Finally, the paper designs and implements the prototypes of syntactic and semantic watermarking, which are used to produce the result of experiment.