Academic journal article JITTA : Journal of Information Technology Theory and Application

Robust Performance Analysis on YAWL Process Models with Advanced Constructs

Academic journal article JITTA : Journal of Information Technology Theory and Application

Robust Performance Analysis on YAWL Process Models with Advanced Constructs

Article excerpt

Abstract

The importance of actively managing and analyzing business processes is acknowledged more than ever in organizations nowadays. Business processes form an essential part of an organization, and their application areas are manifold. Most organizations keep records of various activities that have been carried out for auditing purposes, but they are rarely used for analysis purposes. This paper describes the design and implementation of a process analysis tool that replays, analyzes, and visualizes a variety of performance metrics using a process definition and its execution logs. Performing performance analysis on existing and planned process models offers a great way for organizations to detect bottlenecks within their processes and allow them to make more effective process improvement decisions. Our technique is applied to processes modeled in the YAWL language. Execution logs of process instances are compared against the corresponding YAWL process model and replayed in a robust manner, taking into account any noise in the logs. Finally, performance characteristics, obtained from replaying the log in the model, are projected onto the model.

Keywords: business process management, performance analysis, process mining, Yet Another Workflow Language (YAWL), log replay analysis

INTRODUCTION

Business processes form the heart of every organization, whether small or large, and a number of processes can always be identified within an organization and their information systems. Organizations regularly undertake process improvement activities to ensure that their operational processes are as effective and efficient as possible (Weske 2007). In doing so, identification of both frequently occurring flaws and performance bottlenecks in existing process executions are essential, as they typically become the starting point of any optimization efforts (Bentley and Davis 2009; Persse 2006). Furthermore, carrying out performance analysis on existing and planned process models offers a great way for organizations to detect issues within their processes. For instance, Brataas et al. (1997) presented a framework to measure the performance of workflows that involve both manual and automated activities.

Business process simulation is commonly used to carry out performance analyzes of (TO-BE) processes (Ardhaldjian and Fahner 1994). One of the drawbacks of using process simulation techniques is that the performance analysis results are only as good as the input data that is being used to generate the simulation experiments (i.e., garbage-in-garbage-out). Setting up realistic simulation experiments can be very time-consuming (Reijers and van der Aalst 1999). Simulation requires a model which reflects the behavior of a process, including the data and resource perspectives (Wynn et al. 2008). Traditionally, making such models requires a collaborative effort between key stakeholders and process analysts (Poulymenopoulou et al. 2003), which implies lengthy discussions with workers, extensive document analysis, and careful observation of participants.

Executable process models, i.e., workflows, contain descriptions of what tasks need to be performed, when they need to be performed, by whom they need to be performed, what information they need, and what information they produce. Workflows exist independently of workflow management systems that provide process executions, i.e., a workflow in a system may exist merely as a guideline, without enforcing a process execution description (van der Aalst 1998). A Process-Aware Information System (PAIS) is a software system that manages and executes operational processes involving people, applications, and/or information sources on the basis of process models (Dumas et al. 2005). Most PAISs today provide some kind of event log (also referred to as a transaction log or audittrail) (van der Aalst et al. 2003b), where an event often refers to an activity (within a process instance); each event is logged with descriptors such as timestamp, event-type and executing resource. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.