This paper describes a improved document expansion method in information retrieval. Using clustering expansion, we found amazing feedback based on the ordinary query expansion. The traditional information retrieval model has restrict on search related documents due to it only checks the existence of query terms in documents can not considering the context of documents. Now we retrieve documents by VSM and cluster the top-x documents to re-ranking the result set from the formore. Experiments show that the new method has realized a good advancement comparing with the traditional method.