Process mining is a relative young research area which consists of process modeling and data mining. Process discovery as a part of the process mining focuses on converting event logs into process models. Petri Nets formalism is identified as the most convenient resulting process model. However, it is not entirely satisfying and needs to be improved with the purpose of covering the temporal aspects of the studied system. Compared with it, DEVS has the advantage of explicit and concurrent time and separating model from simulation. The objective of this paper is to specify DEVS model as the resulting process model. Based on the existing Two-Phased Approaches in process mining, a region-based approach with the suitable mapping is designed to convert the transition system directly to DEVS. A study case is presented to implement this approach. This paper is a position paper and it needs to be completed. In addition, a dynamic semantic should be designed to solve some typical representational limitations and a simulation engine should be selected. 1. INTRODUCTION Nowadays, the development of the information systems cannot be separated by the operational process. Moreover, more and more events with abundant process information are recorded. Despite the omnipresence of event data, most of the organizations analyze and build models based on expert assumption rather than quantified collected fact. Thanks to process mining [Van der Aalst, 2011], it provides the methods to discover monitor and improve actual processes by using event data to extract process-related information. Process mining combines process design by model-driven approaches and data mining. It includes three types: process discovery, process conformance and process enhancement. This paper anticipates and lays a cornerstone to make a breakthrough on the process discovery area. The purpose of discovery technique is to extract an event log with using recorded historical information and produce a model for simulation. In the process mining,