The traditional top-k query processing uses single user preference to calculate ranking score, which thus has limitations. In this paper, we studied the problem of top-k query by incorporating multi-user preferences. In order to improve the efficiency of query processing, the initial data set is divided into base relational tables in accord with their attributes. Then, using the existing base relational tables, we select a sub-set of id in the original data set, and use the sub-set to run top-k query. We proposed preprocessing algorithm PPV and PLBA and then proposed the LBA algorithm dealing with top-k queries and proved the correctness and completeness of the algorithm. Experimental results demonstrated that our algorithm improves the query processing in comparison with the original algorithm which runs directly in the original data set.