An Author Authentication Technique For Network-based Textual Document

Natural language watermarking is becoming a hot topic in watermarking technology. Considering the content of text as a semantic integral, it embeds watermarks into the text via the use of existing semantic conversion techniques. This technique can be applied to author authentication of textual document.Wu et.al have proposed two different approaches to natural language watermarking, APS2 (Authorship Proof Scheme 2) and APS3, both appearing more robust than their prior counterparts. APS2 is based on TMR, which makes use of complex semantic analyses and techniques such as inarching, pruning, and replacement etc. Due to the complexity and expression variety in natural languages a lot of efforts must be paid to optimize APS2. Although APS3 is not TMR-based and therefore more concise, it still has space for improvement.An improvement for Wu's approaches is presented in this thesis. When watermarks are to be embedded and extracted, it first uses a replacement word table to select a word and its replacement word in every watermarking sentence. Then these two words will be sent to SHA-1 for digest calculation, the results of which will decide the binary values of both words. Finally the word to be used will be decided by the watermarks. Furthermore, Hamming coding is adopted for testing and rectification in watermarking authentication.A prototype system has been developed using Visual C++ 6.0 and C++ Builder 6.0 in a MS-Windows environment.