The business rules that govern the behaviour of a business process can be hardcoded in different ways in a software application. The modernization or improvement of these applications to a process-oriented perspective implies typically the modification of the business rules. Frequently, legacy systems are not well documented, and almost always the documentation they have is not updated. As a consequence, many times it is necessary to analyze the source code and databases structures to transform them into a business language more understandable by the business experts involved in the modernization process. Database triggers are one of the artefacts in which business rules are hardcoded. The authors focus on this kind of artefact, having in mind to avoid the manual analysis of the triggers by a database expert and bringing it closer to business experts. To achieve this, they need to discover business rules that are hardcoded in triggers and translate them into vocabularies that are commonly used by business experts. In this chapter, the authors propose an ADM-based architecture to discover business rules and rewrite them into a language that can be understood by the business experts.
TopIntroduction
In recent year’s management theory, it has been attached high relevance to a process-oriented perspective on organizational (re)structuring. One of the main reasons for the evolution of information systems in organizations is the need for changing their business processes and requirements. Yet to date, organizations still experience difficulties to adapt their information systems to this process-oriented perspective, especially because it requires them to undergo a modernization process, in which business experts are involved. Unfortunately, it is common to find that the documentation about business rules is outdated and, as a consequence, the only source of information about current business rules is databases and source code. But these technical artefacts are hard to understand by business experts.
Modernization processes typically start with a discovery phase in which technical artefacts, such as source code and databases structures, are analyzed. Triggers or integrity constraints are one of the technical artefacts in which business rules are hardcoded. These represent rules for correct persistent states that a database can take and define the bounds for well-formed transactions that are allowed against database.
The modernization of software applications in a system described by means of a business process model involve two important activities: the definition of the model and the description of the business rules. The process model is described in a language such as Business Process Model and Notation (BPMN) (OMG, 2010b). However, the policies or statements that govern the behaviour of the company need to be described by means of business rules, for example with a language such as Business Vocabulary and Business Rules (SBVR) (OMG, 2008). Business rules can be seen as a common language between the business-side and the IT-side of organizations. The necessity to combine both perspectives (imperative and declarative) has been analyzed in papers such as (Skersys, T. et al., 2012a), (Skersys, T. et al., 2012b). This paper is focuses on the modernization process of the declarative description of the business process by means of business rules. Therefore, the business rules need to be captured from legacy information systems, since often the documentation is outdated or, simply, does not exist. In these cases, several types of sources must be analyzed to discover the business rules. Legacy Information Systems represent a serious problem for software maintenance process (Bisbal J., Lawless D., Wu B., Grimson J., 1999). (Stavru S., Krasteva I., Ilieva S., 2013) analyze the challenges for legacy information systems in the scope of Model Driven Modernization, with an enumeration of the main organizational and technical categories of challenges
One type of source is triggers, which are hardcoded in databases. They describe the relation between data values and are commonly written in proprietary languages, such as PL*SQL or Transact*SQL, which are difficult to understand by a non-database expert. In these cases, it is common to manually translate the source code of triggers to natural language or a business language, in order to make it easier to understand by the business expert. Triggers are, on the one hand, a well-structured knowledge base, linked to tables in databases and, on the other hand, a clear specification of the events that launch actions or methods to be executed. However, the procedural code of triggers is close to programmers but too distant from natural language, and, therefore, far from the language that business experts handle.
To bring these two worlds closer, in this work we consider a model-driven reverse engineering process, in which a set of metamodels at different levels of abstraction are provided. The levels of abstraction are proposed in Architecture Driven Modernization (ADM) (OMG, 2010), that correspond, bottom-up, to Platform-Specific Model (PSM), Platform-Independent Model (PIM) and Computation-Independent Model (CIM). As is analysed in the section of related works, although there are several solutions that analyse the modernization process in business processes, none of them tackling all the steps of the problem.
The paper is organized as follows: