目前文本水印算法多基于文本外在特征,很少利用文本内容上的内在关联性,通过对文本句子的主语和宾语进行指代冗余分析,构造用于嵌入水印的指代冗余矩阵,再根据水印信息和矩阵编码规则确定指代冗余矩阵修改位置,利用实体状态编码和状态转移操作修改原文本完成水印嵌入。该算法可以抵抗格式变换、同义词替换等攻击,具有较低的文本修改率和较好的鲁棒性。
Most of the current text watermarking algorithms are based on external features of text,but seldom the internal relationship of text content is used. This paper constructs the coreference redundancy matrix as the direct carrier by analyzing the coreference redundancy of the sentences' subjects and objects in the text,locates the coreference redundancy matrix elements which need to be modified according to the watermarking information and matrix coding rules,then revises the original text based on the entity status coding and state transfer operations. This algorithm can resist the attacks of the format changing,synonym substitution and so on with a lower modification rate and better robustness.