Home // International Journal On Advances in Software, volume 7, numbers 1 and 2, 2014 // View article
Authors:
Bastian Roth
Matthias Jahn
Stefan Jablonski
Keywords: meta model derivation; meta model inference; derivation of meta model changes; refinement of meta models; language patterns
Abstract:
Designing concise meta models manually is a complex task. Hence, newly proposed approaches were developed, which follow the idea of inferring meta models from given model examples. Unlike most approaches in the state of the art, we accept arbitrary model examples independent of a concrete syntax. The contained entity instances may have assigned values to imaginary attributes (i.e., attributes that are not declared yet). Based on these entity instances and the possessed assignments, a meta model is derived in a direct way. However, this meta model is quite bloated with redundant information. To increase its quality, we provide recommendations for applying so-called language patterns like inheritance or enumerations. For this reason, the applicability of those patterns is analyzed concerning the available information gathered from the underlying model examples. In addition to our previously published work, we also support the derivation of meta model changes based on modifications and extensions of the initial example models. Furthermore, change recommendations are provided wherever possible. This new approach for iteratively building, modifying and refining meta models enables users to focus on the real world instances. Consequently, they are not distracted by keeping the meta level in mind and thus are able to design meta models rapidly.
Pages: 31 to 43
Copyright: Copyright (c) to authors, 2014. Used with permission.
Publication date: June 30, 2014
Published in: journal
ISSN: 1942-2628