Abstract: The semantic Web is increasingly accepted as the next generation of WWW. The foundation of semantic Web is resource description framework (RDF). As more and more information is represented in RDF format, the efficient dissemination and filtering of RDF information becomes an important problem. In information dissemination systems under the semantic Web, input RDF data should be matched with a database of user profiles, which can be represented as RDF graph patterns. Based on the characteristics of RDF and several restrictions on the RDF graph, a novel matching algorithm is proposed for RDF graph patterns. RDF graph and RDF patterns are all traversed from a special node to form BFS trees; the matching state of two BFS trees are represented as an AND-OR tree to avoid backtracking. RDF patterns are indexed according to the concept model to further improve efficiency. Experimental results show that the algorithm is much more efficient than conventional graph matching algorithms.