Web Applications (WAs) have been often used to expose business processes to the users. WA modernization and evolution are complex and time-consuming activities that can be supported by software documentation (e.g., process models). When, as often happens, documentation is missing or is incomplete, documentation recovery and mining represent an important opportunity for reconstructing or completing it. Existing process-mining approaches, however, tend to recover models that are quite complex, rich, and intricate, thus difficult to understand and use for analysts and developers. Model refinement approaches have been presented in the literature to reduce the model complexity and intricateness while preserving the capability of representing the relevant information. In this chapter, the authors summarize approaches to mine first and refine later business process models from existing WAs. In particular, they present two process model refinement approaches: (1) re-modularization and (2) reduction. The authors introduce the techniques and show how to apply them to WAs.
TopIntroduction
In the last decade, Web applications (in short WAs) have been often used to expose business processes to the users. WAs, in fact, are complex applications mainly implemented by means of distributed server-side components (written, e.g., in JSP, Java, Php, Perl) that are exercised by client-side components (written, e.g., in HTML, Ajax, Flash) running on a web-browser user-interface. Web applications are event-based applications in which the user executes the underlying application processes by interacting with the application user-interface.
WAs maintenance, evolution and modernization are complex and time consuming activities that can be supported by software documentation, e.g., business process models underlying the applications, as well as by application and domain knowledge. Unfortunately, often, such documentation and knowledge do not exist or, if they exist, they are not aligned with the actual application implementation because in most of the cases application maintenance and evolution activities are not well-documented. In these situations, documentation recovery and mining techniques can represent an appealing opportunity to reconstruct and complete the missing and inaccurate documentation.
Existing techniques to recover and mine process models (e.g., (van der Aalst et al. 2003), (Zou et al. 2009)), however, tend to generate models that are quite complex, rich and intricate, thus difficult to understand and use for analysts and developers. Such process models are known in the literature as “spaghetti” processes (Veiga et al. 2009).
To limit process model complexity and intricateness (e.g., (Alves et al. 2006), (Mendling et al. 2007)) several approaches have been proposed in the literature. In this chapter, we will present some relevant types of process refinement techniques, that is state-of-the-art techniques that improve process model readability while preserving the capability of the process to represent the relevant information about the application. In particular, we will detail two different approaches to improve the process model readability: (1) process model re-modularization; and (2) process model reduction.
Process model re-modularization (e.g., (Bose et al. 2009), (Di Francescomarino et al. 2010), (Reijers et al. 2008)) consists of improving the recovered process models by increasing their modularity degree. To this aim, process elements are clustered together according to heuristics and then represented in the initial processes through subprocesses (i.e., complex activities).
Process model reduction (e.g., (van der Aalst et al. 2003), (Marchetto et al. 2011), (Marchetto et al. 2011b), (Tomasi 2012)), instead, consists of reducing process model complexity by removing process elements according to heuristics.