Aspect Mining for Dynamic Service Orchestration

Traditional web service orchestration is always implemented by business process work flow or I/O matching. But this method will invoke some problems including too many composite results, inefficiency, and most important the composition is static for service orchestration is based on service concerns, but not taking account of service status. Based on Aspect-Oriented analysis, service concerns are divided into base themes and aspect themes in this paper. Base themes are traditional service entities, while aspect themes are cross-cutting service concerns. Incoming filtering chain and outgoing filtering chain, which are aspect-based expressions, are used to describe service providers or customers status. Service orchestration can be achieved through these two filtering policies dynamically for aspect-based expressions and results are dynamic. Testing cases have been used in Chinese Mobile-Oriented City Information System, and the results demonstrate that dynamic service orchestration can be achieved through two aspect filtering chains.