The emergence of fully-automated Web service composition as a potential facilitator of both eBusiness and ambient or ubiquitous computing is to be welcomed. However this emergence has exposed the need for flexibility and adaptivity due to the fundamentally unreliable nature of the networks and infrastructure on which the component services rely. Furthermore, a key to driving forward acceptance and adoption of this growing set of technologies is the improvement of the user’s overall experience therewith. Our experimentation has proven that it is quite possible to generate inflexible and only partially adaptive service compositions using out of the box A.I. planners. Because modifying the planner is beyond the scope of our research, we seek to use methods of pre-processing and post-analysis to enable AI planners to produce adaptive compositions. In this paper, the current state of our research is presented along with a proposed direction for improving the reconciliation of user needs...
Steffen Higel, David Lewis, Vincent P. Wade