In XML keyword search, the main challenge is to achieve the relevant results to the query and proper ranking of query results. The keyword search approach for XML databases should effectively address 1) search intention of the user, 2) keyword ambiguity problems, 3) result grading problem. Based on the requirements, we propose a keyword search approach XDMA in which two indices are built and a keyword search technique using the indices is presented. Further, by incorporating the concept of dependence of indices and mutual measure, we propose mutual summation and subsequently, define the mutual score to determine desired T-typed node. Also, we propose a formula for similarity measure to find the exact data through the selected T-typed node. Furthermore, the query results having comparable relevance scores are graded. Extensive experimental evaluations have demonstrated the effectiveness of our proposed approach, XDMA.