Solving the Movie Database Case with QVTo

This paper proposes a solution to the movie database case [3] of the Transformation Tool Contest 2014. The objective of the movie database case is to derive a set of performance results that indicate the ability of model transformation languages to process large models with millions of objects. The case study is based on the IMDb movie database that stores information about movies, actors, actresses, and ratings. We use QVT Operational Mappings (QVTo, [4]) for solving the different tasks of the movie database case. QVTo is a textual, imperative model transformation language based on OCL [5] that is standardized by the OMG. It natively supports metamodels specified in EMF [6] such as the provided IMDb metamodel. In this paper, we use the QVTo implementation of the Eclipse Model to Model Transformation (MMT) project1. The Eclipse implementation of the QVTo standard is open source and already widely used in other open-source and academical projects. It is used, for example, within the Graphical Modeling Framework (GMF2) and in the Papyrus project3. Recently, it has been used for translating software design models to verification models [1] and for generating operational behavior specifications out of declarative ones [2]. In our implementation, we created seven transformations for solving the different tasks of the movie database case. We implemented the three main tasks and all of the extension tasks. Our implementation demonstrates that QVTo enables a concise specification of the solutions. Four out of seven tasks require less than 30 lines of code. Our benchmark results show that the Eclipse implementation of QVTo is currently able to handle input models with a few thousand objects in a reasonable amount of time. The paper is structured as follows. We first briefly review the movie database case in Section 2 and QVTo in Section 3. Thereafter, Section 4 describes our solution that we implemented in QVTo. We provide benchmark results concerning runtime of our transformations in Section 5 before concluding the paper in Section 6.