Implementing Business Process Recovery Patterns through QVT Transformations

Abstract

Abstract. Traditionally, software maintenance takes only technical information into account to evolve legacy systems. However, business knowledge, which could help to improve the comprehension of legacy systems, is rarely recovered. Probably, that knowledge is not considered due to the fact that business knowledge recovery is a problem with a non trivial solution. This paper contributes to the solution of this problem through the use of a set of patterns and the implementation through QVT transformations, which takes KDM (Knowledge Discovery Metamodel) models concerning the system and obtains BPMN (Business Processes Model and Notation) models to represent the embedded business knowledge. A case study reports that the transformation obtains cohesive and loosely-coupling business processes diagrams; and it is scalable to large systems. As a consequence, the business processes recovery can enhance the maintenance since they provide the business concept location in legacy systems, among other improvements.

Business process archeology using MARBLE

“…Particularly, the set of patterns comes from the previous work presented in [42,44], although the implementation of these patterns has been substantially improved. For example, the name of the main methods is now transformed into a callable unit in the KDM model by using the name of the main class, since this name is more representative.…”

An empirical comparison of static and dynamic business process mining

“…The pattern recognition and the generation of business process models is also implemented in the developed tool by means of QVT transformations [22]. In addition, the discovered business processes can be refined by business experts through the tool, since it provides graphical model editors for both KDM and business process models.…”

MARBLE. A business process archeology tool

“…When a specific structure is detected in the KDM model at L2, each pattern indicates what elements should be built and how they are interrelated in the business process model at L3 [16]. This pattern matching is implemented through a QVT transformation [17].…”