Abstract: A new keyword search method is proposed to look for the subgraphs which should contain all or part of the query keywords on the data graph. First, we consider two factors, Page Rank and the keyword position, to overcome the weakness of current works which only considered distance relationship between nodes and ignored the similarity between node and query during searching. Second, we design new algorithms for node searching and result generating. The closest relationship between two nodes can be obtained. At last, we verified this method could return results with higher quality.