Research Of Varied Strategy On Chinese Text Zero-Watermark

With the wide application of internet, lots of texts which own copyright can be transmitted easily by internet, so how to protect fusion of unauthorized text has become important research of text's copyright protection. For this problem, this we construct a new watermark scheme and do some deep research on scheme's structure, watermark constructing and testing method. Then main work and research results are following.(1)Based on analysis of the existing watermark scheme and strategy, the formula expression of zero-watermark was given. According the Chinese linguistics' and text's character, verb and sentences' length character are combined together to construct a varied strategy on Chinese text zero-watermark scheme, which has three courses that are pro-process, watermark construction and test.(2)In the course of pre-process, text threshold is brought in so that the division of text will be easily. By text threshold, the text can be divided into two different types, and the corresponding pre-process algorithm is designed.(3)In the course of watermark construction, the function structure is projected and the detailed algorithm of verb and sentences'length is designed. In the Chinese text zero-watermark strategy based on verb character, utilize linguistic technology to choose verb word times, and watermark will be got by combining the verb words with key words of text together. In the Chinese text zero-watermark based on sentences'length which chooses sentences'length as element of watermark, information redundant principle is used to deal the chosen sentences to get the sentences'information table, which table's data will be combined to produce watermark. The performance of watermark is proved by theory.(4)In the course of watermark test, the function structure is projected and the detailed algorithm is designed. In the test of verb character watermark, utilize information fusion to get verb word set and key word set from watermark firstly, and compute the overlap ration of those sets of disputed text, then the overlap ration are used to judge the watermark result. In the test of sentences'length watermark, the text's sentences information table will be fetched by information fusion from watermark, after that watermark judge result will be given by computing the sentences'information correct rate with table and text's content by majority voting method. After that, analyze the complexity of watermark's algorithm.