With the development of web technology, the use of dynamic web pages and the personalization of page contents become more and more popular. Currently, the information of page is protean and the structures of different pages are vastly different, the traditional thinking of web information extraction technology has been difficult to adapt to the situation. In this paper, proposes a web information extraction method based on extended XPath policy through the analysis of structural features of web pages on tourist theme. This algorithm avoids the defects of traditional web information extraction technology; it is simple, practical, high cleaning efficiency, accuracy, and saving the overhead of the system.