A reverse engineering process for inferring data models from spreadsheet-based information systems: An automotive industrial experience

Nowadays Spreadsheet-based Information Systems are widely used in industries to support different phases of their production processes. The intensive employment of Spreadsheets in industry is mainly due to their ease of use that allows the development of Information Systems even by not experienced programmers. The development of such systems is further aided by integrated scripting languages (e.g. Visual Basic for Applications, Libre Office Basic, JavaScript, etc.) that offer features for the implementation of Rapid Application Development processes. Although Spreadsheet-based Information Systems can be developed with a very short time to market, they are usually poorly documented or in some case not documented at all. As a consequence, they are very difficult to be comprehended, maintained or migrated towards other architectures, such as Database Oriented Information Systems or Web Applications. The abstraction of a data model from the source spreadsheet files represents a fundamental activity of the migration process towards different architectures. In our work we present an heuristic- based reverse engineering process for inferring a data model from an Excel based information system. The process is fully automatic and it is based on seven sequential steps. Both the applicability and the effectiveness of the proposed process have been assessed by an experiment we conducted in the automotive industrial context. The process was successfully used to obtain the UML class diagrams representing the conceptual data models of three different Spreadsheet-based Information Systems. The paper presents the results of the experiment and the lessons we learned from it.