Interrogative Agents for Data Modeling

IntroductIon The problem of changes in software development is a complex one, and it is almost impossible to avoid it. Indeed, the continuous evolution of the real world causes frequent changes in functional requirements, which entail frequent modifications to the software, yielding a gradual decay of its overall quality. To tackle this problem, two meth-odologies have been proposed: waterfall method-ologies, and incremental/iterative methodologies. The formers try to prevent changes, whereas the second ones consider system development as a step by step process. The concept of software refactoring is at the base of iterative and incremental methodologies. According to Fowler software refactoring is " … a disciplined technique for restructuring an existing body of code, altering its internal structure without changing its external behavior. " (M. Fowler, http://www.refactoring.com/). Analogously, changes to the database structure are also relatively frequent (Roddick, 1995). They are particularly critical, since they affect not Interrogative Agents for Data Modeling only the data, but also the application programs accessing them (Ambler & Sadalage, 2006; Kara-hasanovic, 2001). Therefore, similarly to software refactoring, database refactoring aims to modify the database schema, and to change the corresponding application programs accordingly. Ambler & Sadalage (2006) gave the definition of database refactoring: It " is a simple change to a database schema that improves its design while retaining both its behavioural and informational semantics. ". The database refactoring is the basis of evolutionary data modeling methodologies, which are the database analogous of the iterative and incremental ones. Ambler & Sadalage (2006) also observed that a disadvantage in the application of refactoring is the lack of mature supporting tools. In this paper we deal with the problem of developing tools supporting the evolutionary data modeling process. First of all, we observe that the characteristics of the problem can be naturally framed in the agent paradigm, because the evolutionary data modeling can be seen as a process in active databases able to change their beliefs and structure. Moreover, the evolutionary data mod-eling can be compared to the design of an agent acting in an open environment: the environment can be represented by the user needs and requirements (which change in an unforeseeable way), while the database development process is represented by the evolution of a reactive agent. Then, by following the AOSE (Agent-Oriented Software Engineering) view, we show that the use of tools and techniques from AI (Artificial Intelligence) can help facing the problem of …