Using Process Mining to Business Process Distribution 1 Faramarz Safi Esfahani Department of Computer Science Islamic Azad University, Najaf Abad Branch Esfahan, Iran. fsafi@fsktm.upm.edu.my Masrah Azrifah Azmi Murad, Md.Nasir Sulaiman, Nur Izura Udzir University of Putra Malaysia Faculty of Computer Science and Information Technology,Selangor, Malaysia. {masrah, nasir, izura}@fsktm.upm.edu.my ABSTRACT Service Oriented Architecture (SOA) is by far the most pervasive architecture which includes several building blocks among which orchestration engine is under special focus. Although, there are a number of centralized orchestration engines to execute business processes described by BPEL language in SOA, you may find several decentralized orchestration engines and their purpose is decomposing a BPEL process to several software agents to improve quality factors such as adaptability, performance and so forth. As these process distribution methods break a BPEL process to its building activities and encapsulate each activity in one agent, it results in producing a lot of agents whose interactions and resource usage would degrade the run-time environment. This paper proposes an intelligent process distribution (IPD) based on a process mining approach in which the selection of activities that should be encapsulated in agents, depends on the previous behavior of process instances. The recommended IPD approach will improve three aspects of system quality. First; is the amelioration of business process adaptability with run-time environment, second; choosing the best agent granularity based on detecting most relevant activities and encapsulating them in agents and third; is decreasing of resource usage due to reduced and improved number of produced agents and messages. Furthermore, we proved our method using a mathematical approach. Keywords: Adaptive systems, Business Process Mining, BPEL, Service Oriented Architecture, Mobile Agents, Workflow, Distributed Orchestrate Engine. 1. Introduction In service oriented architecture, business processes are executed by an orchestrate engine that is responsible for running the activities of a process. Normally, a single engine is applied to manage a business process and scalability is satisfied by replicating orchestration engines which do not obviate the problems of centralized engines. On one hand, many researchers are working on BPEL business process distribution. The main idea is distribution of activities of a BPEL process among some autonomous agents or sub processes interacting through a middleware. On the other hand, a number of researches have been focusing on the idea of process mining to extract useful information from process log files. The mined information will be used to detect most relevant parts of a business process, drawing run-time Petri net model of a business process and discovering social networks which are extremely important to provide more adaptable business processes with run-time environment. In this paper we are going to reduce resource usage and improve the adaptability of business processes considering the execution history of previous executed processes. Accordingly, the contributions of this paper are: 1) Designing an intelligent method based on process mining to ameliorate the granularity of agents at compile time. 2) Improving the adaptability of the BPEL processes considering execution history of previous executed agents using process mining. 3) Reducing the number of agents and exchanged messages compared to the traditional methods of BPEL distribution. 2. Background and Related Work BPEL: The Business Process Execution Language or BPEL briefly supports web services relationships and interactions in business transactions, message exchange correlation for long running message exchanges, parallel processing of activities, the mapping of data between partner interactions, consistent exception and recovery handling [1, 2]. BPEL activities [1, 2] can be classified as basic activities that perform some primitive operations and structured activities that define control flow. The key BPEL basic activities are Invoke, Receive, Reply, Assign, Compensate, Compensate-Scope, Empty, Exit, Throw, Rethrow, Validate and Wait whereas structured BPEL activities are Flow, For-Each, If, Pick, Repeat-Until, Scope, Sequence and While, respectively. Process Mining: Service Oriented Architecture contains a variety of events that can be logged. In addition, log data can be used for process mining purposes, its goal is to build models without apriori knowledge, based on sequences of events, one can look for the presence or absence of certain patterns and deduce some process models from it[3]. In [4] a framework for an agile mining of business processes introduced. In this framework, analyzing 1 Faramarz Safi is a PhD Candidate in Faculty of Computer Science and Information Technology, University of Putra Malaysia (UPM), Selangor, Malaysia. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. SAC’09, March 8-12, 2009, Honolulu, Hawaii, U.S.A. Copyright 2009 ACM 978-1-60558-166-8/09/03…$5.00.