In recent years, the network and information technology is more and more ripe we always search for data or files through the Internet. The resources on the World Wide Web are increased day by day. The amount of information exchange and speed of update are also grown. Many methods depend on the match of vocabularies between information request and searched objects. For example, we usually adopt some keywords and Boolean operators to form the query to search the Internet for interested information. Unfortunately, users do not always use proper words and operators to form the query for the search. The result is related or unrelated information is both retrieved.The retrieval method becomes critical for us to get more accurate results. In order to improve the searching process, a RDF-based mechanism integrated with word sense disambiguation technique is proposed to semantically index and retrieve web pages on the World Wide Web. The approach is to describe the resources by RDF(S) metadata and store them. The proposed method also has additional advantage of the ability of further integrated into the Semantic Web Service.