The current tendency of information system organizations to work with process models has led to an increase in the importance of the comparison of these models stored in some collection or repository. Comparing process models for similarity or dissimilarity search faces two main challenges: (1) Finding the most relevant/irrelevant process model to a particular process or query of interest, and (2) Decreasing the time complexity to achieve this task. Many approaches have been proposed to improve the relevance of returned processes depending on matching the graphs that represent these processes. This generally turns to a graph matching problem known to be an NP-Hard problem. In this paper, we propose to decompose process model graphs into modules in order to improve the precision of the returned results while keeping a polynomial time complexity of matchmaking. The Modules are sub-parts of the process model that capture the most important sub-structures of the process graph. With these modules, the original process graph can be represented by a tree which is more easier to match. We have evaluated our approach within a benchmark of process graphs. The results of our experiments confirm that our approach obtains more relevant results with less time than a typical graph matching based method.