A new method on the detection of near-replicas of web pages

Near-replicas of web pages have seriously decreased the efficiency of search engine (SE). In this paper, we present a new method to detect near-replicas of web pages. Firstly, the styles of text structures in web pages are analyzed and classified; then according to the styles of the text, different methods are used to get the text structure, which will be represented as a matrix; Finally, the similarity will be calculated by extracting the features dynamically from the matrix. Experiments show that this method can not only improve the computing efficiency but also ensure high precision and recall.