Reuse of proven process models can increase modeling efficient by avoiding reinvent the wheel and ensure the quality of process models. To provide a better support for reuse, the retrieve mechanism of a process repository should be able to propose similar process models that ranked according to their similarity degrees to users for reuse by modification or customization. As a process model and a query model can both be viewed as rooted, directed, and acyclic graphs, the problem of querying structural information of BPEL processes has reduced to a graph matchmaking problem. In this paper we present a novel and efficient graph-based algorithm for querying structural information of BPEL processes based on an inexact matchmaking semantics. Our algorithm performs in the worst case in polynomial time in the orders of the query graph and the process graph.