A Chinese Text Zero-Watermark Algorithm Based on Statistical Features
YJ Meng, MW Liu, WJ Liu, Weiting Chen · Lanzhou University Institutional Repository · 2014
In order to protect the copyright of Chinese digital texts, contraposing the shortcomings of existing watermark algorithms like poor anti-attacking and robustness, this paper proposed a text zero-watermark scheme based on statistical features with computational linguistics as reference, built the construction and detection model for text's zero-watermark, and also designed the core algorithm of the scheme. In this scheme, textual features including feature word set, short sentence percentage, average sentence length and part-of-speech statistical property were extracted first, then were fused and encrypted as the zero-watermark of the protected text. For purpose of validating whether our watermark scheme is effective and feasible or not, simulation tests were provided. The experimental results indicated that our method has better performance of robustness and anti-attacking, especially against the attack of syntax or format, etc.