—Aiming at the limitation of existing Zero- Watermark algorithm in the copyright protection, a Chinese text Zero-Watermark scenario based on threedimensional space model was proposed. In the scenario, the three-dimensional model was constructed by using the features of high frequency words, sentence length and its weights. Afterwards, the algorithm generates a watermark based on the text abstract-set which can be extracted later by comparing the distance of each sentence point with the centre of model. In case of any copyright conflict, the extraction algorithm extracts the abstract-set from the copyright disputes text again, and by calculating the semantic distance to define original copyright owner. Furthermore, the effectiveness and feasibility of the proposed algorithm was proved with algorithm simulation. The simulation results also showed that the algorithm had good robustness, especially for syntactic transfer, synonyms replacement and shear attacks occurring randomly.
展开▼
