A Robust Text Zero-watermarking Algorithm based on Dependency Parsing

Abstract

A new robust text zero-watermarking algorithm is proposed, which extracts features and generates watermark from text. The algorithm specially aims at Chinese text documents. Using Dependency Parsing, key words are extracted from the sentences including Negative Words. Then these words are encoded by Tongyici Cilin and encrypted. The encrypted string combined timestamp and author information is the generated watermark and sent to be registered in Certification Authority. When detecting watermark, we calculate the similarity between the watermark of disputed text and its watermark registered in CA. Experimental results show that the algorithm has good robustness.