In this paper, we have presented MoDAR, a model-driven approach for developing dynamically adaptive BPEL-based
service-oriented systems using aspects and rules. We have introduced the methodology of separating the variable part from
the stable part of a process and modeling the variable part as business rules, as well as weaving these rules into the stable
base process. We have also developed a platform using a number of the state-of-the-art technologies that support the
model-driven development and execution of rule-weaved BPEL processes. The targeting system is dynamically adaptable in
the sense that rules as the variable part of a system can change at runtime without affecting the base process. Experimental
results have shown that systems developed and generated with MoDAR have slight performance overhead but significantly
simplifying the development process and achieving runtime adaptability.
Our future work will focus on tackling several limitations identified during the real-life medical clinic case study. In particular,
we will enhance the re-usability of rules in the development tool, and investigate runtime validation techniques. We
plan to develop new techniques on top of the MoDAR platform to facilitate the development of context-aware applications.
Furthermore, we also plan to translate MoDAR to a formal model to improve the expressiveness of the MoDAR models, and
to support the property verification of MoDAR applications.