Research On Chinese Text Zero-watermark Based On Syntax And Semantic

Copyright authe nticatio n is an important techno logy of copyright protection. Itcan identify the authors of the carriers after some routine mod ificatio ns, and detectthe tampered carriers. As one of the most important copyright authe nticationtechno logies, digita l watermarking e mbeds authe ntication informatio n whic h is usedfor identifying the copyright owner into digita l carriers.The text watermark ing reviews are ana lyzed and tradiona l te xt watermarkingmethods have the shortcoming of changing te xt semantic in this paper.Zero-watermark methods, which do not introduce any changes to carrie r, can solve theproble m. The presented text zero-watermark methods ma inly focus on t he statistica lfeatures of characters, Chinese Characters, words or sentences. However, textsemantic is the most important informatio n of text. Feature extraction fro m textsemantic can greatly resist the edittings of keeping text semantic, which is suitable forcopyright authentication.To extract features o f text se mantic, some natural langua ge processingtechno logies are e mplo yed in this paper. A new zero-watermark ing algorithm based onDependency Pars ing co mbined with re le vant la nguage k nowledge is proposed toextract syntax feature words in Chapter3. A HowNet based method of semanticreleva ncy computatio n between words and a zero-watermark ing a lgorithm based onlexica l cha in are proposed in Chapter4. This method of sema ntic releva ncycomp utation between words proposed is usd for extracting feature words fro m lexica lchain. To increase the security a nd robustness, the extracted feature words areencoded by To ngyic i Cilin and encrypted with Logistic c haotic mapping.Experime nta l results s how that the a lgorithms are robust to routine modifications andfragile to malic io us tampering, whic h can meet the demand for copyrightauthenticatio n.