Finding Structure and Characteristics of Web Documents for Classification

Many Web documents containing the same type of information , would have similar structure. In this paper, we examine the problem of nding the structure of web documents and present a hierarchical structure to represent the relation among text data in the web documents. Due to the loose standard of web page publishing, diierent authors can use diierent wordings (labels) to label the same information. We introduced a labels discovery algorithm that uses the hierarchical structure extracted from the web pages. The algorithm discovers similar labels which describe the same kind of information. Such labels would help us nd the structure of the web documents. Experiments have shown that the algorithm can successfully discover similar labels and the structure obtained by our method can distinguish web pages accurately.