Nowadays, data dissemination often involves online databases that are hidden behind query forms, thus forming the deep web. Lately, there has been a lot of research interest on supporting query answering over the deep web. To answer a deep web query efficiently, the current approaches generate a query plan for each query independently. However, in practice, deep web queries issued by a user over a short period of time can often share similarities. This, if properly utilized, can help us in generating more efficient query plan. In this paper, we have developed a solution for generating query plan for a deep web query based on the similarities between a given query and a set of earlier queries. Our algorithm systematically finds the reusable components of earlier query plans, and then develops a new query plan reusing these. While the resulting query plans may not be optimal, they are likely to enable more data reuse, and hence, speedup the execution.