UML2 context: uma extensão da UML para modelagem de contexto

Abstract

Researchers and industry professionals recognize the importance of developing context-aware systems, as these systems increasingly are part of everyday life of people. To design, communicate design decisions and represent the structural and behavioral aspects of these systems, developers commonly use the Unified Modeling Language (UML), which is widely recognized as the standard language of software modeling. However, given the increasing diversity of modeling techniques and UML's inability to represent the concepts of context-aware systems, is particularly challenging for developers to design and objectively communicate design decisions of such systems. Usually developers don’t have the tools to facilitate the representation of context in an appropriate manner, whereas propose notations to solve the problem, further aggravating the problem of heterogeneity of current modeling techniques. The main problem is that UML is inaccurate to represent the concept of context and rigid to incorporate new concepts. Another problem is that academic and commercial tools of current modeling - including IBM RSA, Astah, Borland Together - do not support the representation and validation of context models. This paper therefore proposes an extension of UML for modeling context, which aims not only to allow the representation and validation of the key aspects of context, but also enhance communication of project decision of context aware systems. It also proposes a domain specific modeling environment that enable the modeling of context-aware systems, which was implemented as an Eclipse platform plugin using the GMF frameworks, EMF and UML2tool. This environment will not only represent the concept of context following the proposed approach and to evaluate the correctness of the models. The extent and the proposed tool was evaluated through a questionnaire addressing the proposed model compared with modelings that using the standard UML. This avaliation allowed us to evaluate the real benefits of work and where the results suggest that the modularization of context information in a new concept called UML2Context brings benefits compared with the decomposition of such information with the pure UML. The results link that UML2Context increased the rate of correct answers in 28.41%, reduced the effort of interpretation in 61,03% and improved the interpretation of context models in 35.98%, proving effective for modeling ubiquitous systems.